University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Colorado School of Mines
|
Figure 3.13: AMR Star Ground
readings unaffected, yet low enough to block the noise from the golf cart motor.
With the wiring and power distribution completed, the AMR is fully capable of
teleoperation. The last section outlines the safety controls developed as the teleoper
ation capabilities were tested.
3.2.11 Engineering Safety Controls
Safety was paramount throughout the AMR’s design and construction; autonomous
vehicles can be a serious danger, especially during preliminary testing, resulting in
damaged toolboxes, equipment, property, or injured persons. Thus, numerous safety
measures were implemented during development.
First, there are two emergency off (EMO) switches located at the front and rear
of the vehicle. The robotic controller detects these switches in a high-priority inter
rupt service routine and immediately shuts off throttle and engages the brake when
the EMO switch is depressed. The autonomous controller can also send a software
emergency stop, which behaves similarly to the physical EMO switches.
Second, when changing to remote mode (see Section 3.2.6), the robotic controller
engages the brake and blocks throttle and steering commands until it detects a specific
startup procedure from the operator. To unlock the robotic controller, the operator
must, while the brake is engaged on the transmitter, toggle the reverse switch on the
47
|
Colorado School of Mines
|
CHAPTER 4
LOCALIZATION ALGORITHM DEVELOPMENT
This chapter covers the equations and algorithms involved in localization. To
gether, the Ackermann vehicle dynamics, the matching algorithm, and the extended
Kalman filter form the overall localization algorithm. This localization algorithm is
based on the scanning laser sensor research efforts in order to determine if these al
gorithms can cope with the inherent problems of ultrasonic sensors. Where problems
arise, new or modified algorithms are proposed.
Below is an overview of the localization algorithm. This chapter elaborates on
each step.
1. Obtain an environment map of the mine wall. The assumption is that an
accurate a-priori map is not provided, so the environment map is built from an
initial pass of the mine using dead reckoning data.
2. Run the localization algorithm. The localization algorithm is based on the
ICP-EKF algorithm developed by Madhavan et al. [19].
(a) Initialize the vehicle.
(b) Perform the EKE prediction step, which uses Ackermann vehicle dynamics
to predict the vehicle pose.
(c) Perform the EKE measurement update, which corrects the dead reckoning
drift using the sonar range readings in an iterative matching algorithm.
(d) Iterate the algorithm while the vehicle is moving.
The list above is re-visited in more detail at the end of the chapter.
4.1 Ackermann Vehicle Dynamics
Similar to most automobiles, the AMR has Ackermann steering; Ackermann ve
hicle dynamics mathematically describe a means of localization via dead reckoning:
51
|
Colorado School of Mines
|
Vx = S ■ cos(^), (4.1)
%, = S.sm(0), (4.2)
— = — • S' • tan (a) [3], (4.3)
where Vx and Vy are the horizontal and vertical components of the vehicle speed, S;
6 is the heading of the vehicle, defined as positive in the counter-clockwise direction;
a is the steering angle, also defined as positve when turning counter-clockwise; and
L is the vehicle wheelbase. Vx and Vy can be integrated to give position, (X, Y).
Figure 4.1 is a graphical illustration of these variables.
X
Figure 4.1: Ackermann Vehicle Dynamics Variables
Ackermann steering dynamics estimate the entire vehicle as a point at location {X, Y )
oriented in direction. 0, analogous to a ‘virtual’ wheel located directly between the
front wheels. The location of (Ah Y) on the vehicle is arbitrarily defined to coincide
with’the ‘virtual’ wheel.
Equations (4.1) through (4.3) are integrated in the discrete time domain. At each
time iteration (k, or every 250 ms), Equations (4.4), (4.5), and (4.6) compute X, V,
and 0, respectively:
52
|
Colorado School of Mines
|
The AMR drives an arbitrary path, with an ending location as shown in Figure 4.3.
Frame {U} remains fixed relative to the earth, while the vehicle frame, {V} travels
with the vehicle. The distance and direction that the AMR traveled is represented
by the vector uI>vorg, with orientation, 6. Note that 6 is defined positively in the
counter-clockwise direction; a negative angle is a sign convention to denote a clockwise
direction. The ultrasonic sensor’s coordinate frame, {S} is fixed relative to the vehicle,
though translated and rotated relative to Frame {V}.
Figure 4.3: Wall Mapping Scenario AMR Finishing Point
To illustrate the coordinate system transformation equations, see Figure 4.4. This
particular illustration uses sonar 8 as an example, but the process is identical for the
remaining range sensors. At time /c, Sonar 8 detects an object some distance away.
Since the orientation of Sonar 8 is known, the range measurement can be converted
into a vector, denoted SP. The same point on the wall can be represented with
vectors VP, which is relative to the vehicle’s coordinate frame ({V}), and VP, which
is relative to the universal coordinate frame ({U}). Equation (4.7) relates SP and
1 P, and by extension, Equation (4.8) relates all three vectors:
|
Colorado School of Mines
|
Figure 4.4: Wall Mapping Scenario: Sonar 8 Range Measurement
"P ii k, m = VKt1k,
•
V p 1fc, m Sp 1fc, m (4.8)
where k is the time iteration index and m is the ultrasonic sensor number index. Note
that the ‘ V is a placeholder for proper matrix dimension agreement. The homogeneous
transformation matrix, "T, in Equation (4.8) consists of 2 components: a rotation
matrix and a translation vector, as shown in Equation (4.9) [5]:
cos 0k - sin 0k 0
U pjd VORG, k 1 sin 0k cos 0k 0 n
(4.9)
0 0 0 1 0 0 1
Zk
0 0 0 1
where 6 is the angle relative to the initial orientation, as depicted in Figure 4.3, and
X and Y are the horizontal and vertical displacements of the vehicle, respectively.
The third dimension, Z, can be ignored for this particular application, but is included
for completeness. Similarly, is broken down in Equation (4.10):
COS/3m - Sin 0m 0
y rp _ y S I& xm VPsORG,m sin 0m COS 0m 0 Urn
(4.10)
s 1m — 0 0 0 1 0 0 1
Zm
0 0 0 1
55
|
Colorado School of Mines
|
where (5 is the angle of the ultrasonic sensor relative to the vehicle’s coordinate frame
({V}) and (x, y) refers to the location of the sonar sensor on the AMR (see Fig
ure 4.5). Again, the variable z is not necessary for a ground vehicle, but is included
for completeness. Each ultrasonic sensor has unique values for x, y, and j3\ a table of
these values can be found in Section A. 11 of the Appendix.
{V}
AMR
SORG, 8 — j" -----
03
Figure 4.5: sorg and ft for Ultrasonic Sensor 8
The overall idea of the map building equations is to build one large matrix (called
‘Map’) that contains the coordinates of every wall measurement in the environment,
using one sensor at a time; for each time iteration &, the sonar sensor’s readings in
Frame {U} give the (x, y) coordinates of one point on the map. Repeating the process
for the remaining sensors gives 13 more sets of (x, y) coordinates. Conglomerating
the 14 points from each time iteration (every 250 ms) during the entire experiment
gives the environment map of the wall. For an example of a resulting environment
map, see Figure 5.11 on Page 79. Additionally, a step-by-step outline of the map
building procedure is given at the end of the chapter. The next section attempts to
find unique features in the environment map (the large matrix called ‘Map’).
56
|
Colorado School of Mines
|
4.3 Matching Algorithm
For this project, the purpose of a matching algorithm is to use exteroceptive
sensor data to identify unique features on a map produced from Section 4.2 and give a
corresponding vehicle pose estimate at each time iteration, k. The pose estimates from
the matching algorithm serve as the measurements for the measurement update in the
Extended Kalman Filter algorithm (Section 4.4). Based on previous research efforts
[4] [1] [19], two different types of matching algorithms are tested: an Iterative Closest
Point (ICP) algorithm and an Iterative Closest Line Segment (ICLS) algorithm.
The fundamental difference between the ICP algorithm and the ICLS algorithm
is how the data association problem is addressed. The ICP algorithm, as its name
implies, simply associates each measured point with the closest point in the environ
ment map. The ICLS algorithm addresses the space in-between points on the map,
associating each measured point with the closest point on the nearest line segment in
the environment map. Figure 4.6 illustrates the difference between the ICP and the
ICLS association method.
Wall Map
O Sensor Reading
O ICLS Association
O ICP Association
Figure 4.6: ICP Versus ICLS Algorithm Data Association
If Point J is a sensor’s measurement of the wall, the ICP algorithm would associate
point E with the measured point, and the ICLS algorithm would associate point M
with the measured point. The coordinates of point M are found using Equation
57
|
Colorado School of Mines
|
(4.11), given by,
EC EJ
Pm = P k + (4.11)
IIECII
where Pm is a vector containing the coordinates of Point M, Pg is a vector containing
the coordinates of Point E, EC is a vector starting at Point E and ending at Point C,
EJ is a vector starting at point E and ending at point J, (EC • EJ) is the dot product
of the vectors EC and EJ, and | |EC|| is the length of vector EC. In general, the ICLS
data association is closer to the original measurements; however, the ICLS algorithm
is computationally more expensive. Repeating the association process illustrated in
Figure 4.6 for each of the 10 wall-profiling sensors builds the data association matrix,
uPwaii (a [10 x 2] matrix).
After data association is complete, the remaining steps are the same for both ICP
and ICLS algorithms. Let q be the matching algorithm iteration number, and A be
a [10 x 3] matrix as shown in Equation (4.12):
I/pT
*9,2
C/pT
q, 4
[/pT
9) 5
E/pT
*9,6
E/pT
Aq = *9,8 (4.12)
E/pT
*9,10
E/pT
*9,11
E/pT
9,12
E/pT
9,13
E/pT
9,14
The matrix A represents the readings from the 10 wall-profiling sensors, expressed
58
|
Colorado School of Mines
|
in the universal coordinate frame, along with a column of ones, which serve as a
placeholder. Note that the numeric subscripts correspond to the sensor number (Fig
ure 3.5) and that only the wall-profiling sensors are used in this analysis (the soft
bumper sensors, i.e. Sonars 1, 3, 7, and 9, are skipped in the numerical ordering
in matrix A). The vector uV^r rn in Equation (4.12) is the transpose of the vector
computed in Equation (4.8). u~Pq,m is a function of the vehicle pose, so it is updated
at every matching algorithm iteration, ç, with the latest values for X match, Ymatchi
and Omatch- Note that the sensor range measurements, 5,Pm in Equation (4.8), are
constant during the ICP or ICLS iterations.
Additionally, let u'PWaii represent the associated points on the wall, corresponding
to either the ICP or the ICLS algorithm. Equation (4.13) calculates the rotation
matrix (R) and translation vector (B) required to align the ultrasonic sensor readings
with UPWall-
(4.13)
Note that the rotation matrix and translation vector, R and B, are similar to yR and
U¥ vorg from Section 4.2, respectively, but they are not identical.
The matrix A is not square, so inverting it requires a pseudo inverse, such as
singular value decomposition. With the rotation matrix and translation vector (R
and B) in hand, there are numerous ways to back out the vehicle displacement and
translation errors, Realizing that R is actually the transpose of
drift)-
yR (from Equation (4.9) on Page 55) makes computing 9drift straightforward:
@drift, q = SHI 1(Bg(l, 2)), (4.14)
where R(l, 2) is the element of R located on the first row and the second column. The
rotation matrix is with respect to the universal frame, {U}, so the rotation affects
the translation vector. Equation (4.15) isolates Xdrift and Ydnft from B:
59
|
Colorado School of Mines
|
[Xdrift,q, Ydrift,q] = BqRq 1 = BqRq. (4.15)
Then, Equation (4.16) updates the matching algorithm vehicle pose, X match, Ymatch
and Omatch, by adding the computed vehicle drift:
Xmatch, q+1 Xmatch, q Xdrift, q
Ymatch, g+1 = Ymatch, q + Ydrift, q
Omatch, q+1 Omatch, q 0drift, q
On the first iteration, the vehicle pose is initialized to the Kalman filter a-priori es
timate (Equation (4.18) on Page 62). The index, g, is incremented, and the process
repeats, starting with updating the data association, until convergence or until a maxi
mum number of iterations is reached. On the last iteration in g, X match,q, Ymatch, q, and
omatchq become the measurement in the Kalman filter ([Xmatch,k, Ymatch,k, 0match,k]. in
Equation (4.28) on Page 64) for the time iteration, k.
After convergence, the last step of the matching algorithm is to compute the
measurement uncertainty. There are many factors that go into the measurement un
certainty, including the confidence in the map building process, the ultrasonic sensor
readings, and dead reckoning drift. One method for addressing the measurement er
ror is to rely on dead reckoning pose for short distances. Adapted from Kolter et ah,
Equation (4.17) computes the standard deviation of the matching results [15]:
^Xmatch, k {Xmatch, qfinal Xmatch, o)
match, k = {Ymatch, qfinal Ymatch, 0 ) (4.17)
_ ®@match, k {Omatch, qfinal Omatch, o)
where arid are the standard deviations of ATwWt, EAoWi, and
Omatch, respectively; qfinal is the value of q at the final ICP (or ICLS) iteration;
and X matcK0, Ymatch,t), and Omatch,o form the initial vehicle pose, defined to be the
Kalman filter a-priori update. The standard deviation in Equation (4.17) reflects
the idea that the vehicle is unlikely to drift far during one time iteration, so the
60
|
Colorado School of Mines
|
farther away the matching algorithm pose deviates from the dead reckoning pose, the
higher the uncertainty. With the variance in vehicle pose defined in this manner, the
Kalman filter is unlikely to change the dead reckoning pose dramatically during one
time iteration; however, small corrections can add up significantly over the course of
testing.
With the vehicle dynamics equations, a technique for map building, and the ability
to localize relative to the map (i.e. the matching algorithm) in hand, all of the individ
ual ‘ingredients’ are prepared for creating the main entrée; the next section develops
an Extended Kalman Filter, which is a method for blending the three ‘ingredients’
together.
4.4 Extended Kalman Filter (EKF)
An EKF provides a means of fusing exteroceptive information from a matching
algorithm with proprioceptive information from dead reckoning data. Combining the
two complementary localization techniques eliminates the problems of using either
algorithm individually. The EKF breaks the process into two steps: a prediction step
and a measurement update. During the prediction step, Ackermann dynamics pro
vide a pose estimate based strictly on odometry and steering angle sensor readings.
The measurement update then uses ultrasonic range measurements and a matching
algorithm to correct for drift. The process repeats at each time iteration. The follow
ing EKF equations are a modified version of the ICP-EKF algorithm from Madhavan
et al. [19].
4.4.1 EKF Prediction Step •
Beginning with the prediction step, Equations (4.4) through (4.6) update the
vehicle pose. Additionally, two states are added to keep track of vehicle slip (ds)
and skid (as) defined as the error in vehicle odometers and the error in in vehicle
61
|
Colorado School of Mines
|
cos(^-i) o 0 0
sin(0fc_i) 0 0 0
t tan(a/c H-û!s^ -i) t_ L1_ cosd 2d (f ac+ fcr +fs a, sfc ,f- ci -i' 0 0 (4.26)
0 0 1 0
0 0 0 1
rr2 0 0 0
ddk
0 < 0 0
Qk — 0 0 <j 2 0 (4.27)
dds, k
0 0 0 (J2
as, k
4.4.2 EKF Measurement Update
The second step of the EKF is the measurement update. At each time iteration,
the matching algorithm uses range readings to measure the wall’s profile and return
a prediction of the vehicle’s location. The results from the matching algorithm serve
as the measurement that fine-tunes the vehicle pose from the prediction step. The
measurement update begins with computing the innovation, yk:
Xmatch, k
Yk = — Hxk, (4.28)
Ymatch, k
@match, k
where H is the measurement sensitivity matrix, defined by Equation (4.29); and
Xmatch, Y-match, and 6match are the output of the matching algorithm (Section 4.3 on
Page 57). The matrix H is defined as follows:
1 0 0 0 0
H = 0 1 0 0 0 (4.29)
0 0 1 0 0
The measurement update continues with computing the Kalman gain, K:
Kk = Pk HT(HPk HT + R average,k) ^ (4.30)
64
|
Colorado School of Mines
|
where
R average
is the covariance matrix of the observational uncertainty, averaged over
a small time window (Equations (4.31) and (4.32)). The term Rfl (for ‘covariance
matrix filter length’ ) is intentionally left as a variable because it can be tuned for best
performance. This technique for producing the covariance matrix of the observational
uncertainty comes from Kolter et al. [15]:
, Rfi-i
(4.31)
'average,
match, k
(4.32)
match, k match, k match, k match, k
match, k
Also known as the a-posteriori update (xj), Equation (4.33) adjusts the vehicle pose
and the slip and skid variables:
Xk
Yk
(4.33)
^s, k
The measurement update finishes by computing the a-posteriori estimate of the co-
variance matrix, P£:
P+ = P," - KkHP;. (4.34)
At the end of the measurement update, the time index, k, is incremented, and the
EKF process repeats, starting with the prediction step, Equation (4.18). The process
continues until the algorithm reaches the end of the collected data.
4.5 Localization Algorithm Summary
Returning to the itemized list at the beginning of the chapter, the steps below
outline the entire ICP-EKF localization algorithm in detail.
65
|
Colorado School of Mines
|
1. Build an environment map:
(a) Define the origin of the universal coordinate frame as the vehicle’s starting
point.
(b) Start the vehicle
(c) At time fc, record the vehicle’s sensor data: the odometers on each wheel,
the steering angle sensor, and the 14 range measurements from the sonar
sensors.
(d) Using the odometry and steering angle values, apply the Ackermann vehi
cle dynamics equations (Equations (4.1) through (4.3)) to determine the
vehicle’s location ((XwMcZe,%,e/iWe)) and orientation (^eMcZe)-
(e) Using the sonar range data and the vehicle pose (location and orientation),
apply Equation (4.8) to find ^Pfc,m for each sensor. This equation trans
lates the sonar range readings to a vector.(a set of coordinates) referenced
in the universal coordinate frame. This step produces 10 measurements of
the wall.
(f) Add the 10 sets of coordinates from the previous step to the environment
map.
(g) Repeat Step 1(c) through 1(f) until map completion (the end of the ex
periment).
2. Apply the localization algorithm to any subsequent path within the environment
map:
(a) Initialize the vehicle’s location relative to the environment map.
(b) Begin moving.
(c) At time /c, record the vehicle’s sensor data.
(d) Perform the EKF prediction step. This step uses the odometry and steering
angle data.
66
|
Colorado School of Mines
|
i. Update the states with Equation (4.18). This step produces an esti
mate of the vehicle pose using dead reckoning.
ii. Calculate P^T, the a-priori estimate of the covariance matrix, using
Equation (4.24).
(e) Perform the EKF measurement update. Using the sonar range measure
ments in a matching algorithm, this step attempts to correct the vehicle’s
dead reckoning drift using the surrounding wall’s texture,
i. Use the sonar range data in either the ICP or ICLS matching algorithm
to correct the dead reckoning drift. The matching algorithm serves as
the Kalman filter ‘measurement.’ This step has several parts:
A. Record the vehicle pose at the start of the matching algorithm,
(*o, 0o)-
. B. Initialize the matching algorithm’s pose estimate, {Xmatch, Ymatch, @match),
with the initial pose estimate.
C. For each sonar range measurement, find the corresponding closest
point on the environment map of the wall (or the closest point on
the closest line segment for the ICLS algorithm). This produces
Hawaii, a [10x2] data association vector.
D. For each sonar sensor, translate the sensor range measurements to
the universal coordinate frame with Equation (4.8), using the lat
est matching algorithm vehicle pose, (Xmatch, Ymatch, 0match)- The
translated vectors form matrix A in Equation (4.12).
E. Using the data association vector and matrix A, find the estimated
dead reckoning drift, (XdrifUYdrift,Qdrift), using Equations (4.13),
(4.14), and (4.15).
F. Use the estimated dead reckoning drift to update the matching
algorithm vehicle pose estimate (Equation (4.16))
67
|
Colorado School of Mines
|
G. Iterate the matching algorithm (Steps C. through F.) until conver
gence or a maximum number of iterations is reached. The iteration
index for the matching algorithm is denoted as q throughout the
chapter.
H. After convergence, compute the standard deviation estimate for
Xmatchi Ymatch•> and Omatch (Equation (4.17)).
ii. Compute the a-posteriori estimate of the states (Equation (4.33)).
This equation updates the dead reckoning pose with observations from
the matching algorithm; in other words, the dead reckoning drift is cor
rected using sonar range readings. In this regard, Equation (4.33) also
produces estimates for the vehicle’s slip and skid (the dead reckoning
drift), which are used in the EKF prediction step on the next iteration.
iii. Compute the a-posteriori covariance matrix estimate (Equation (4.34))
(f) Return to step 2(c) and re-iterate. The iteration index for the ICP-EKF
or the ICLS-EKF algorithm (also known as the time index) is denoted as
k.
The next step is to test the ICP-EKF and the ICLS-EKF localization algorithms.
Using both simulated sensor data and measured sensor data collected at the Edgar
Mine, the following chapter contains the test results.
68
|
Colorado School of Mines
|
CHAPTER 5
RESULTS
This chapter covers the results of the localization algorithm. Initially, data is
fabricated to test the algorithm. After the algorithm is proven to work using simulated
data, it is applied to real data collected at the Edgar Mine. The localization algorithm
and data post-processing is executed in Math Works® MATLAB software.
The preliminary testing is performed exclusively in software, beginning with sim
ulated sensor data. In this chapter, the first section describes the simulator that
produces the sensor data. The ICP-EKF localization algorithm is initially tested in
an ideal situation using the simulated data (Section 5.2). Afterwards, the localization
algorithm is tested in Section 5.3 using data collected from one section of the Army
Tunnel at the Edgar Mine. For this test, dead reckoning drift is added to the original
data to determine whether the localization algorithm can compensate. Due to sensor
calibration issues, a ‘multi-map’ method was created to get the localization algorithm
to work properly. The last test was the ‘double-pass’ test using sensor data from two
full passes of the Edgar Mine’s Army Tunnel (Section 5.4); the localization algorithm
uses the second pass to localize against the first. Because the sensor data was too
noisy for the localization algorithm, the experiment was repeated using simulated
sensor data.
5.1 Simulator
The simulator provides a means of testing the localization algorithms without the
complexity of actual data. The first step is to create an arbitrary vehicle path with
simlulated odometer and steering angle vectors. The wall is created by sending a
vector of sonar data through the wall mapping equations (Equation (4.8) on Page 55)
for Sonar 2 and Sonar 4, while running the simulated odometer and steering angle
69
|
Colorado School of Mines
|
values through the Ackerman dynamics equations (Equations (4.4) through (4.6) on
Page 53). The vector of sonar data begins with a series of line segments, and then
noise is added and the vector is passed through a low-pass moving average filter. The
output is a discrete set of horizontal and vertical displacement values, X sirn and YSim,
that form the vehicle path and a discrete set of points that form a map of the wall.
Figure 5.1 is the resulting map.
Map Simulator
Mine Wall
Vehicle Path
B
>
-5
X (meters)
Figure 5.1: Map Building and Vehicle Path Simulation
After the wall is created, for each point on the vehicle path, [Xsim^ , Vsh^ d, a
simulated set of sonar sensor readings is created. Figure 5.2 is an example of the
simulated sonar data collection process. The sonar projection is a line segment be
tween a range reading of 0 meters and a range reading of 6 meters for each sonar
sensor (simulating a sensor beam). The distance from the sonar sensor to the nearest
intersection point becomes the sonar range reading.
Localization algorithm testing initially occurs using the simulated data. The orig
inal simulated vehicle path is the ‘ground truth.’ A small amount of noise or drift
is added to the odometer data, the steering angle data, and the sonar sensor data.
The localization algorithm from the previous chapter receives the noisy sensor data
and attempts to predict the original ‘ground truth.’ For comparison, the same noisy
70
|
Colorado School of Mines
|
data runs through the Ackermann vehicle dynamics equations to produce the dead
reckoning predicted path. The closer a path is to the ground truth, the better the
performance.
Sonar Range Reading Simulation
-2- Wall Map
' Sonar Projection
O intersection
0 1
-3 -2 ■1 2 3 4 5 6
X (meters)
Figure 5.2: Simulated Range Data Collection
5.2 Localization Algorithm Tests Using Simulated Data
A set of data for the 10 wall-profiling sensors, the odometer, the string poten
tiometer, and the mine wall is fabricated in order to test the localization algorithm.
Furthermore, each sensor has artificial noise added: the odometer has a 30% drift,
the steering measurements from the string potentiometer have a random gaussian
noise with a 10° standard deviation, and the ultrasonic sensors’ range readings are
given a gaussian noise with a 15.2 cm standard deviation. Additionally, at each time
step, the orientation of each ultrasonic sensor, /3m, is given a gaussian noise with a 5°
standard deviation, corresponding to the beam width of the ultrasonic sensors. These
noise values were initial estimates of data confidence and drift in a mine. Figure 5.3
is the resulting graph of the simulated data test using the ICP algorithm.
The ‘ground truth’ line in Figure 5.3 is the vehicle path before adding the noise.
The dead reckoning path calculates the vehicle path using Ackermann vehicle dy-
71
|
Colorado School of Mines
|
ICP Localization Algorithm
"■ Mine Wall
" Ground Truth
- — - EKF Predicted Path
Dead Reckoning
Predicted Path
Î
E
-2-
-6 -4 -2
X (meters)
Figure 5.3: ICP Localization Algorithm Test Using Simulated Data
namics with the noisy odometer and string potentiometer data (measured steering
angle or heading), and the EKF predicted path uses the same noisy data, with the
added benefit of fusing the noisy dead reckoning data with the simulated noisy ul
trasonic range data. Even for a worst-case scenario, the EKF predicted path is able
to eliminate the majority of the dead reckoning drift (‘noise’). The ICLS algorithm
performed in a similar manner—Figure 5.4 shows the results of the same test and
data using the ICLS algorithm.
The ‘ground truth’ line in Figure 5.3 is the vehicle path before adding the noise.
The dead reckoning path calculates the vehicle path using Ackermann vehicle dy
namics with the noisy odometer and string potentiometer data (measured steering
angle or heading), and the EKF predicted path uses the same noisy data, with the
added benefit of fusing the noisy dead reckoning data with the simulated noisy ul
trasonic range data. Even for a worst-case scenario, the EKF predicted path is able
to eliminate the majority of the dead reckoning drift (‘noise’). The ICLS algorithm
performed in a similar manner—Figure 5.4 shows the results of the same test and
data using the ICLS algorithm.
The ICLS algorithm yields identical results as the ICP algorithm, primarily be-
72
|
Colorado School of Mines
|
cause the map of the mine wall has small spacing between points. The only noticeable
difference between the ICLS algorithm and the ICP algorithm is that the tests take
68.3 and 4.2 seconds to run, respectively. In order to run in realtime, the ICLS code
would require optimization.
One method for increasing the processing speed of both matching algorithms is
to simply restrict the map available to the matching algorithm; map points over 8
meters away from the vehicle are an unlikely target as the maximum range reading
for these ultrasonic sensors is only 6 meters. Restricting the map access allows for
very large environment maps without a significant increase in processing time; data
storage becomes the primary limiting factor of map size.
ICLS Localization Algorithm
Mine Wall
Ground Truth
EKF Predicted Path
, Dead Reckoning
Predicted Path
Î
E
>
-4
-6 -4 -2
X (meters)
Figure 5.4: ICLS Localization Algorithm Test Using Simulated Data
Regardless of the algorithm used, the localization algorithm is very robust to noisy
and drifting sensor data. In fact, both the EKF-ICP and the EKF-ICLS localization
algorithms managed to reject the dead reckoning drift when the noise on the steering
and odometry sensors were increased by 50%.
73
|
Colorado School of Mines
|
5.3 Localization Algorithm Test Using Collected Data
With the promising results using simulated data, the next step is to test the
localization algorithms on actual data. The first step is to build a map—Figure 5.5
uses Equation (4.8) to build the environment map of a portion of the Edgar Mine
Army Tunnel for each sonar sensor from data collected on March 11, 2010. Figure 5.5
highlights several issues with the ultrasonic sensors in a mine: the range readings are
very noisy, some sensors (Sonars 5 and 8) appear to be incorrectly calibrated, and
the diagonal sensors (Sonar 5, 6, 11, and 12) have large spikes in their range readings
that appear to be more than noise.
Map Building Sonar 2
Sonar 4
Sonar 5
Sonar 6
Sonar 8
Sonar 10
Sonar 11
Sonar 12
Sonar 13
Sonar 14
5 10
X (meters)
Figure 5.5: Map Building Using Ultrasonic Sensor Range Data
First, regarding the noise issue, when the AMR was stationary, the standard de
viation of some sensors were over one meter, though the average standard deviation
came out to a more-reasonable 0.25 meters. Regarding the calibration, these partic
ular sensors seemed to drift from day to day, so they required frequent calibration.
Regarding the large spikes in the diagonal sensors, the ultrasonic sensors exhibited
similar behavior, though more severe, when ranging a flat wall at a steep angle of
incidence (see Section 3.2.4 on 35).
The environment map in Figure 5.5 is very cluttered. Thus, for aesthetic purposes,
74
|
Colorado School of Mines
|
an averaged mine wall was created. Figure 5.6 compares the environment map (the
raw data) to the averaged wall; the averaged wall is the result of vertically averaging
the mine wall measurements from the environment map. The averaged mine wall is
not the environment map, but rather, a cleaned-up representation of the environment
map, so it is used in the remaining figures in this section. The noise in the environment
map far exceeds the wall’s texture; the texture seen in the averaged wall is an artificial
by-product of the averaging technique.
Averaged Map Building
• Environment Map
Averaged Wall
0)
0)
E
>
0
-5 5 10 15 20
X (meters)
Figure 5.6: The Averaged Wall
The drifting calibration for individual sensors were a persistent problem in all
data collected. To alleviate the issue, a wall is constructed for each wall-profiling
sensor (instead of building one composite wall using all sensors), making a total of 10
maps; thus, each line in Figure 5.5 represents one map. The matching algorithm only
associates a sensor’s measurements with points from its own map. Other techniques
include averaging the points together to build one composite, but this technique tends
to dull the unique mine features into one continuous line, making localization nearly
impossible. Figure 5.7 are the results of the localization algorithm tests using the
75
|
Colorado School of Mines
|
multi-map method.
ICP Localization Algorithm
Mine Wall
Ground Truth
- - - EKF Predicted Path
Dead Reckoning
Predicted Path
0)
0)
E
>
-2-
-10 -5 20 25
X (meters)
Figure 5.7: ICP-EKF Algorithm Test Using Collected Data
To reiterate, the mine wall in Figure 5.7 is the result of vertically averaging the
mine wall measurements from the environment map in Figure 5.5, and the matching
algorithm uses the environment map (now comprised of 10 maps, one for each wall-
profiling sonar sensor). For the purpose of testing, the ground truth is the vehicle
path predicted from the original dead reckoning data. A 20% drift was added to the
original odometry data, and a Gaussian noise with a 2° standard deviation was added
to the steering angle. The noisy sensor data was used to form the dead reckoning
predicted path in Figure 5.7. Notice that the EKF predicted path performed little
correction in the odometry data (the paths are about the same length); with no
unique features, the localization algorithm is unable to determine an exact position.
However, the localization algorithm was able to correct vehicle orientation, so the
EKF predicted path was, for the most part, aligned with the ground truth path.
Furthermore, as Bakumbu predicted [1], the localization algorithm easily degenerates
with the relatively flat walls of this particular section of the Army Tunnel, as shown
in Figure 5.8.
Degeneration occurs when the matching algorithm causes the pose estimate to
76
|
Colorado School of Mines
|
ICP Localization Algorithm
— Mine Wall
Ground Truth
- - - EKF Predicted Path
Dead Reckoning
Predicted Path
E
>
-4
-10 -5
X (meters)
Figure 5.8: ICP-EKF Algorithm Degeneration
get off-track. Once off-track, the range readings have little correlation with the
environment, so the matching algorithm is unable to converge, or converges on the
wrong wall segment, causing the pose estimate to become even more off-track. This
cascading effect results an unstable localization algorithm, which causes the vehicle
pose oscillation depicted in the EKF predicted path of Figure 5.8. As Bakumbu
suggests, matching algorithm results are ignored in these portions of the mine [Ij.
Repeating the same test for the ICLS algorithm gave the results in Figure 5.9.
Similar to the ICP localization algorithm, the ICLS localization algorithm was able
to correct the vehicle orientation, but not the odometer readings.
This section provides evidence that the ICP-EKF algorithm can work given per
fectly consistent range measurements; if the range measurements do not change during
subsequent passes, the localization algorithm demonstrates a solid ability to elimi
nate some of the dead reckoning drift. The following section tests the localization
algorithm in a more realistic scenario; the experiment involves two separate passes
of the Edgar mine to determine whether the localization algorithm can use the data
collected during the second pass to localize against the first.
77
|
Colorado School of Mines
|
ICLS Localization Algorithm
, _ , _ , Dead Reckoning
2 6
s;
>
o
-6
0
5 10 15 20
X (meters)
Figure 5.9: ICLS-EKF Algorithm Test Using Collected Data
5.4 Double Pass Testing
The final localization algorithm test uses data from two separate passes of the
Edgar Mine Army Tunnel. Using dead reckoning, data from the first pass builds
the environment map. Similar to the technique used by Makela [20], the localization
algorithm receives sensor data from the second pass to localize relative to the first data
set’s environment map. The purpose of this test is to determine if ultrasonic range
data can localize relative to a map created by ultrasonic range data. This would
prove useful if the AMR would need to return to a previously traversed location.
Figure 5.10 is the map built from the first pass, before and after noise correction.
Environment Map, Pass #1 Environment Map, Pass #1
Raw Data 4 Noise Corrected
5
6
8
10
11
12
13 Sonar 13
5 14
E
> >
0 0
10 20 30 40 50 10 20 30 40 50
X (meters) X (meters)
(a) Before Noise Correction (b) After Noise Correction
Figure 5.10: Environment Map of the Edgar Mine Army Tunnel Using Collected Data
78
|
Colorado School of Mines
|
The environment map is built using dead reckoning data; thus, dead reckoning
drift is inherent. Makela et al. corrected for the drift using a linear scaling factor
on the odometry data relative to a known passage distance [20]. However, since an
emergency rescue vehicle is not intended to be a surveying instrument, an absolutely
accurate map is unnecessary, so drift correction, or some other means of ground truth
(such as a total station), for the environment map was not implemented.
A second complete pass of the Army Tunnel forms the test data set. The starting
position is identical for both passes. Figure 5.11 is an overlay of the two data sets,
created from dead reckoning and ultrasonic range data. The test data set has a
noticeable drift compared to the environment map, which expected for dead reckoning.
Even with the noise correction for both data sets, the mine wall thickness (a measure
of confidence) is on the order of one meter, which is relatively large.
Edgar Mine Environment Map Overlay
Environment Map (Pass #1)
Test Data (Pass # 2)
-10 0 10 20 30 40 50 60
X (meters)
Figure 5.11: Edgar Mine Environment Map Overlay
Any attempt to localize using the map and test data set depicted in Figure 5.11
proved futile. Averaging techniques, the multi-map technique (each sonar sensor can
only be associated with its own data), and the single map technique (each sonar sensor
can be associated with any data), all resulted in either no dead reckoning correction
or an unstable localization algorithm. The localization algorithm relies on consistent
79
|
Colorado School of Mines
|
range measurements, which the ultrasonic sensors are unable to provide; the uncer
tainty and noise of the ultrasonic sensors washes out the unique features of the mine
wall, essential erasing its ‘fingerprints.’ The multi-map method employed in the pre
vious section appeared to correct the vehicle orientation because the uncertainty in
the environment map was essentially ignore. Once factoring in the uncertainty of
the map-building process, the uncertainty in the matching algorithm far exceeds the
uncertainty in dead reckoning for any reasonable amount of distance traveled. Thus,
the matching algorithm is unable to correct the dead reckoning drift.
The inconsistent range readings caused the difficulties experienced when using
actual data. To test this theory, the simulator is revisited. Figure 5.12 is an overlay
of two simulated data sets. Similar to Figure 5.11, the test data set has a slight dead
reckoning drift and each sonar sensor has range and beam width noise to duplicate
sensor response in the mine. Unlike the actual range data, the simulated sonar data
has repeatable, Gaussian noise about the wall.
Simulator Environment Map Overlay
• Environment Map (Pass #1)
Test Data (Pass # 2)_______
-10
-5 20 25
X (meters)
Figure 5.12: Simulator Environment Map Overlay
The two simulated sets of data get the same treatment as the actual data sets.
The first pass is used to create the environment map and the localization algorithm
is tested using the second set of data. Figure 5.13 are the results. With the simulated
80
|
Colorado School of Mines
|
data, the ICP localization algorithm has little trouble with localization on the second
pass; the AMR could return to any point on its map with a fair amount of accuracy.
Thus, ultrasonic range finders could theoretically pass the double-pass test.
ICP Localization Algorithm
Environment Map
Ground Truth
EKF Predicted Path
Dead Reckoning
Predicted Path
-2 0 2 4 6 8 10 12 14 16
X (meters)
Figure 5.13: ICP Localization Algorithm Double Pass Test Using Simulated Data
The failure of the localization algorithm on the collected data and its success on
the simulated data is due to one fundamental difference: the simulated data set is
intentionally formulated with wall features that exceed the sonar range measurement
noise. The sonar range measurement noise has been a recurring issue throughout
the project; the confidence in the range measurements increases dramatically as the
sensor approaches the critical angle (Section 3.2.4 on Page 35) and the sensors were
unable to pick up wall texture in the collected data (Figure 5.11). Furthermore, the
four sensors mounted at an angle greater than the critical angle exacerbate the noise
issue, as their standard deviation far exceeds the assumed standard deviation (as seen
in Sonars 5 and 6 in Figure 5.10(a)). In other words, the sensor noise washes out
the texture in the Edgar mine wall. Thus, the ultrasonic sensors are incapable of the
resolution required for localization at the Edgar Mine.
81
|
Colorado School of Mines
|
CHAPTER 6
CONCLUSION
The primary goal of this thesis was to produce a localization algorithm using ultra
sonic sensors, along with the documentation required to expand upon MineSENTRY
experiments. To demonstrate the capabilities of localization algorithm, the first step
was to build a map; though the ultrasonic sensor readings were noisy and inconsis
tent; a map of the Edgar Mine Army Tunnel was successfully built (Figure 5.10).
Using the map-building algorithm, a simulator was developed to test the theoretical
capabilities of a matching algorithm (Section 5.1). Lastly, an ICP-EKF localization
algorithm and an ICLS-EKF localization algorithm were developed (Sections 4.3 and
4.4) and successfully tested on simulated data (Section 5.4). Though designed for
use with scanning laser sensors, computer simulations suggest that, using ultrasonic
sensors, the ICP-EKF (or ICLS-EKF) localization algorithm consistently improves
upon the dead reckoning predicted path, allowing accurate localization in the thick
smoke common in mine disaster scenarios.
The matching algorithms were also tested on range data collected in the Edgar
Mine (Section 5.4); however, the sensors’ noise exceeded the variation in the mine
wall, preventing the localization algorithm from working properly. This does not
imply that ultrasonic sensors are insufficient for localization, but rather suggests that
further testing is needed using alternative sensors.
6.1 Recommendations for Future Work
An immediate next step is to examine alternatives to the particular ultrasonic sen
sors used, and to test the alternatives in a mine environment. Though the ultrasonic
sensors’ wide detection pattern is advantageous in some situations (e.g. robustness
to vehicle tilting, object detection), a narrow beam width is required for wall profil-
|
Colorado School of Mines
|
ing because of the lower uncertainty. Alternatively, one research project developed
a sonar system that directly addressed the wide beam width problem inherent in
most ultrasonic range finders [16]. Inspired by successful usage of ultrasonic waves
for navigation in nature (e.g. bats and dolphins), Kreczmar developed a sonar system
with one transmitter and two receivers. Similar to the two ears on a bat. the extra
receiver aids in triangulating the source of a reflected acoustic wave, as shown in
Figure 6.1. The graph on the left (Figure 6.1a) illustrates the range measurements
from a single sonar as the sensor is rotated through the environment. The sensor
is unable to determine the direction of the reflecting acoustic wave, resulting in the
sweeping arcs. The graph on the right (Figure 6.15) illustrates the performance of the
“Tri-auler” sonar system; the added receiver allows for better triangulation of objects
within the environment, eliminating the sweeping arcs. Implemented on the AMR.
the “Tri-auler" sonar system may reduce the sensor uncertainty and noise due to the
inherent wide beam width of ‘traditionaF ultrasonic range finders, thereby improving
the performance of the localization algorithm.
b)
4.0 rr ..."I-----T..."-T... 1-----[
3.5 F -
3.0 F -
1 '
2.5 F —
:
2.0 F
Y(m
u 1.5 F □ +
sonar 1.0 F sonar U..
0.5 F -
0 h '------ ------------------------§r r
i : i 1—1 W
-0.5 L-L
0 1 0 1 2 3 4 5 6
X[n%l
Figure 6.1: “Tri-auler’" Sonar System [16]
Additionally, the sensors used suffered from a low critical angle: the diagonally-
oriented sensors had difficulty detecting the mine wall. Ultimately, the alternative
sensors need to be tested for their critical angle and re-oriented accordingly for future
iterations of the MineSENTRY project.
84
|
Colorado School of Mines
|
Another recommendation for future work involves converting the localization al
gorithm into a full SLAM algorithm. The current navigation routine follows along the
main adit using a wall-following algorithm with logic to avoid the side drifts. This
works well if the main adit is relatively straight with no branches, or if a mine map is
readily available to anticipate the side drifts. However, the environment map is not
always available, so to be truly autonomous, a vehicle requires a SLAM algorithm.
A successful localization algorithm is a step towards a full SLAM algorithm; the ICP
(or ICLS) matching algorithm used in localization plays intricate role in the SLAM
algorithm of many research efforts. The main difference between a SLAM algorithm
and the navigation routine currently implemented is a SLAM algorithm’s ability to
correct dead reckoning drift on the first pass (the mapping stage). This is necessary
if the mine is ‘circular’ (i.e. paths overlap), as the robotic vehicle would need to
accurately recognize a previously-traversed path during its forward progress. The
• advantages of a SLAM algorithm are clear; a full SLAM algorithm allows a robotic
vehicle to navigate mine passages without a-priori knowledge of the environment and
with no human intervention, while simultaneously correcting dead-reckoning drift.
The navigational abilities would be superior to the simplistic wall-following tech
niques currently implemented.
6.2 Implications of Research
Even without a full SLAM algorithm, the AMR could still prove to be an in
valuable tool for emergency rescue workers. Equipped with a simple wall-following
control algorithm and the wireless communication control algorithms developed in
work done by Meehan [22], the AMR can autonomously establish and maintain a
wireless mesh network for vital rescue operation communications. Further equipped
with the navigation routines developed in work done by Hulbert [13] and the ICP-
EKF algorithm developed in this thesis, the AMR has the added capability of re
85
|
Colorado School of Mines
|
A.4 Robotic Controller to Servo Controller Communication
Description: The communication between the Robotic Controller and the motor
control board is determined by the motor control board (Roboteq AX3500) command
set and configuration. The robotic controller initiates all communications (except the
AX3500 Watchdog) and the AX3500 responds in one of two ways depending on the
type of command sent by the robotic controller. The AX3500 command set consists
of ASCII strings which correspond to either a request for an action or a request for
data. If a command is issued to request an action, the AX3500 will reply with a
plus (+) to acknowledge the request. If a command is issued that requests data, the
AX3500 will reply with the data to acknowledge the request. If any error occurs with
the communication, the AX3500 will reply with a minus (-) to indicate the command
should be repeated.
All numbers either sent to or received from the AX3500 are represented in hexadec
imal format with ASCII characters, and all commands are followed by the carriage
return character, “\r.”
The data is transferred via standard RS-232 signal levels at a 9600 Baud, syn
chronous data stream with 7 data bits, 1 start bit, 1 stop bit, and even parity. Refer
to Table A.2 for the packet structure specifics.
Table A.2: Robotic Controller to Servo Controller Communication Packets
Byte # Name Value Type Unit Description
0:5 Brake Position “!Bnn\r” or “!bnn\r’; String N/A Brake position relative to zero,
where ‘zero is about half of the full
desired travel, “nn” is a number in
hexadecimal between 00' and 7F (0-
127). In this case, !B7F\r causes a
full release of the brake and !b7F\r
causes the brake to be fully engaged.
6:10 Steering Position “!Ann\r” or “!ann\r” String degrees Steering position in terms of angular
position of the wheels, nn is a num
ber in hexadecimal between 00 and
7F (0-127). In this case, !A7F!bnn\r
causes the steering wheel to turn
right at full speed and !a7F!bnn\r
causes the steering wheel to turn left
at full speed.
96
|
Colorado School of Mines
|
A.5 Robotic Controller to Autonomous Controller Communication
Description: The communication between the Autonomous Controller (AC) and
the robotic controller (VC for vehicle controller)) is determined by the robotic con
troller mode set and status. The AC initiates all communications (except emergency
brakes) and the VC responds in one of two ways depending on the type of packet sent
by the AC. The AC communication set are ASCII strings which correspond to either
a request for an action or a request for data. If a command is issued to request an
action, the VC will reply with a sensory response and a corresponding packet-type to
acknowledge the request. If a query is issued that requests data, the VC will reply
with a sensory response to acknowledge the request. If any error occurs with the
communication, the VC will reply with...
All numbers either sent to or received from the AC are represented in hexadecimal
format with ASCII characters and all commands are followed by the carriage return
character “\r.”
The data is transfered via standard RS-232 signal levels at a 115200 Baud, syn
chronous data stream with 8 data bits, 1 start bit, 1 stop bit, and no parity.
Table A. 3: Robotic Controller to Autonomous Controller Header (common for all
packets)
Byte # Narpe Value Format Unit Description
0:7 Start Sequence “{start:}’" string N/A The start sequence is used in case one
side is not in sync with the other.
8 Packet Type 0 uint 8 N/A This field indicates that the packet is
a system packet.
9 Length X uint 8 N/A Length of the packet in bytes, exclud
ing the start sequence, packet type,
and length fields.
97
|
Colorado School of Mines
|
Table A.4: Robotic Controller to Autonomous Controller System Status Packet
Byte # Name Value Format Unit Description
0:7 Start Sequence “{start:}” string N/A The start sequence is used in case one
side is not in sync with the other.
8 Packet Type 0 uint 8 N/A This field indicates that the packet is
a system packet.
9 Length X uint 8 N/A Length of the packet in bytes, exclud
ing the start sequence, packet type,
and length fields.
0x00: Manual Control
10 Mode See description uint 8 N/A 0x01: Radio Control
0x10: Autonomous Control
0x00/0x01: Brake engaged
0x00/0x02: Softbump Front
0x00/0x04: Softbump Rear
0x00/0x08: SC Reset
11:12 Status See description uint 8 N/A 0x00/0x10: Invalid RC Signal
0x00/0x20: SC non-responsive
0x00/0x40: Vehicle too slow
0x00/0x80: FWD/Reverse
0x01/0x00: Throttle saturation
Table A.5: Robotic Controller to Autonomous Controller Sensor Response Packet
Byte # Name Value Format Unit Description
0:7 Start Sequence “{start:}” string N/A The start sequence is used in case one
side is not in sync with the other.
8 Packet Type 0 uint 8 N/A This field indicates that the packet is
a system packet.
9 Length X uint 8 N/A Length of the packet in bytes, exclud
ing the start sequence, packet type,
and length fields.
10:12 Response Fields OxXXXXXX uint 8 N/A Which fields hold data
x:28 Sonar 0—6000 uint 16 ■mm 14 sonar readings
x:8 IR 0—6000 uint 16 mm 4 IR Readings
x:4 HE 0—255 ■uint 8 dm/s 4 velocity readings
x:2 Odometer 0—65535 uint 16 m Odometer reading
1 BP -60—60 int 8 deg String Pot Steering Angle
1 Throttle 0—100 uint 8 N/A Throttle Setting (x %)
1 Brake 0—100 uint 8 N/A Brake Setting (x %')
0x00: Manual" Control
10 Mode See description uint 8 N/A 0x01: Radio Control
0x10: Autonomous Control
0x00/0x01: Brake engaged
0x00/0x02: Softbump Front
0x00/0x04: Softbump Rear
0x00/0x08: SC Reset
11:12 Status See description uint 8 N/A 0x00/0x10: Invalid RC Signal
0x00/0x20: SC non-responsive
0x00/0x40: Vehicle too slow
0x00/0x80: FWD/Reverse
0x01/0x00: Throttle saturation
98
|
Colorado School of Mines
|
Table A.6: Robotic Controller to Autonomous Controller Emergency Packet
Byte # Size Name Value Format Unit Description
0:7 2 Start Sequence “{start:}” string N/A The start sequence is used in case one
side is not in sync with the other.
2 1 Packet Type OxFF uint 8 N/A This field indicates that the packet is
a query packet.
3 1 Length 1 uint 8 N/A length of the packet, excluding the
start sequence, packet type, and length
fields.
1 Emergency See uint 8 N/A 0x00: Remove Estop
4
Stop Description 0x01: Set E-stop
A.6 Autonomous Controller to Robotic Controller Communication
Table A.7: Autonomous Controller to Robotic Controller Header (common for all
packets)
Byte # Name Value Format Unit Description
0:7 Start Sequence “{start:}” string N/A The start sequence is used in case one
side is not in sync with the other.
8 Packet Type 0 uint 8 N/A This field indicates that the packet is
a system packet.
9 Length ' X uint 8 N/A Length of the packet in bytes, exclud
ing the start sequence, packet type,
and length fields.
Table A.8: Autonomous Controller to Robotic Controller System Startup Packet
Byte # Name Value Format Unit Description
0:7 Start Sequence “{start:}” ■ string N/A The start sequence is used in case one
side is not in sync with the other.
8 Packet Type OxFF uint 8 N/A This field indicates that the packet is
a query packet.
System Status See uint 8 N/A 0x00: Request System Status
10
Request Description 0x01—OxFF: Future Use
99
|
Colorado School of Mines
|
Table A.9: Autonomous Controller to Robotic Controller Query Packet
Byte # Name Value Format Unit Description
0:7 Start Sequence “{start:}” string N/A The start sequence is used in case one
side is not in sync with the other.
8 Packet Type OxFF uint 8 N/A This field indicates that the packet is
a query packet.
9 Length 3 uint 8 N/A Length of the packet, excluding the
start sequence, packet type, and length
fields.
0:13—Sonars
14:17—IR
18:21—HE (Velocity)
22——Odometer
10:12 Sensor Query See uint 8 N/A 23—SP (Steering Angle)
Type Description 24—Throttle Setting
25—Brake Setting
26—Mode
27—Status Bits
Table A.10: Autonomous Controller to Robotic Controller Command Packet
Byte # Name Value Format Unit Description
0:7 Start Sequence “{start:}” string N/A The start sequence is used in case one
side is not in sync with the other.
8 Packet Type OxFF uint 8 N/A This field indicates that the packet is
a query packet.
9 Length 3 uint 8 N/A Length of the packet, excluding the
start sequence, packet type, and length
fields.
10 Requested -60—60 uint 8 deg Position command that is imple
Steering Posi mented immediately in position mode,
tion and is queued along with the speed in
speed mode.
Requested See uint 8 N/A 0x00: Forward
11
Direction Description 0x01: Reverse
12 Requested 2—100 uint 8 dm/s Throttle command that requests a ve
Throttle Value . locity setting.
13 Requested 0—100 uint 8 N/A Brake command that requests.a veloc
Brake Setting ity setting.
Soft-Bump See uint 8 N/A 0x00: Soft-bump enabled
14
Disable Description uint 8 N/A 0x01: Soft-bump disabled
15 Soft-Bump 0—100 uint 8 N/A Setting that controls the sensitivity of
Sensitivity the soft-bump response, 0 indicating a
very low sensitivity and 100 indicating
a very high sensitivity.
0x00/0x00/0x00: Send no data
16:18 Sensor Query See Description uint 8 N/A OxXX/OxXX/OxXX: Send Requested
Data
0x3F/OxFF / OxFF : Send all data
100
|
Colorado School of Mines
|
ABSTRACT
During the last decade, the Mine Safety and Health Administration (MSHA) has
reported a significant number of accidents related to off-highway dump trucks,
accidents that have cost a number of human lives and millions of dollars in equipment
and lost production.
Between 1990 and 1998, there were 133 accidents involving 23 fatalities as a
result of collisions of off-highway trucks with other objects-vehicles, or people in open-
pit mines. In 1998 alone, 13 fatalities occurred in metal/nonmetal and open-pit coal
mines when off-highway trucks ran over smaller vehicles or people not visible to the
truck operator.
The objective of this dissertation is to develop a real-time software system using a
combination of a global positioning system (GPS), wireless communications
networking, and 3D mapping technologies. This system, VirtualMine, is designed to
improve the safety conditions of dumping tasks and collision warning in open-pit
mining operations. It calls for a reliable GPS-based computer guidance system to be
loaded into a panel computer mounted in the vehicle cabin. The system automatically
warns the driver of nearby hazards - e.g., the dumpsite’s edge or another vehicle.
This dissertation presents the results of tests carried out at Colorado School of
Mines survey field and at Morenci Mine, Arizona, Tests focused on GPS accuracy,
vehicle tracking, on-demand 3D-contouring, proximity warning, and wireless data
transfer.
The tests at Morenci have demonstrated that VirtualMine can display truck position
with respect to the edge of the dump as well as with respect to a remote vehicle in
real-time. The results of this dissertation show that the position of haul trucks with
respect to topographic landmarks and other vehicles can be displayed in real-time
with sub-meter accuracy.
|
Colorado School of Mines
|
ACKNOWLEDGMENTS
I would like to express my profound gratitude to my advisor, Dr. Kadri Dagdelen, for
his support and guidance in accomplishing this Ph.D. dissertation.
I am very grateful to all my committee members: Dr. Willy Hereman, Dr. Mark
Kuchta, Dr. Levent Ozdemir, and my committee chairman Dr. John Steele.
I would like to show my appreciation for Ing. Jaime Lomelin, Ing. Octavio Alvidrez,
the National Council of Science and Technology (CONACyT), and to the University of
Guanajuato, who supported me in this program.
I am very thankful to my parents, Antonio Nieto-Antunez and Maria Del Carmen, for
their support and affection and for showing me the importance of knowledge, which
inspired me to pursue this dream.
This dissertation is in honor of my grandfather Antonio Nieto-Vargas., who I am
sure, would have been very proud of this accomplishment, and of my grandmother,
Maria Eugenia.
I dedicate this thesis to my wife Karina who was deeply involved in helping me to
accomplish this beautiful and intensive Ph.D. program, and to my children: Kary,
Jacqueline, Valentina, and Antonio.
ix
|
Colorado School of Mines
|
1
CHAPTER I
1 INTRODUCTION
1.1 System Description
During the last decade, the Mine Safety and Health Administration (MSHA) has
reported a significant number of accidents related to off-highway dump trucks. These
accidents resulted in the loss of several human lives and millions of dollars in
equipment and lost production costs.
Each year, powered haulage used in mining operations is involved in hundreds of
accidents resulting in over 20 deaths, primarily when a truck backs up over a smaller
vehicle or backs over the edge berm of a waste dump. In both types of accidents, the
truck operator’s inability to see dangers in the truck’s blind spots, behind and to the
side of the truck, has been a major contributing factor.
The objective of this dissertation is to develop a software system using the Global
Positioning System (GPS), wireless communications networking, and 3D mapping
technologies to reduce the number of such accidents, and improve working conditions
in open-pit mining operations.
To achieve this goal, the following objectives must be met:
• The system must be able of track the vehicle in real-time with respect to the
edge of a dumpsite, as well as to other vehicles, and to warn the driver of
close proximity or collision.
• The system must be user friendly and must be easily understood by the
driver, relying on simple and representational computer graphics.
|
Colorado School of Mines
|
2
• The system must be capable of sub-meter accuracy regarding the vehicle
position with respect to the safety berm and to other vehicles.
• The vehicle’s position must be shared between all operating vehicles, using
a wireless radio network.
• The system must be able to update the mine’s geometry on demand, based
on remote vehicle coordinates.
• The system must be tested in a mining operation to verify operability,
effectiveness, and operator acceptance.
1.2 Summary of Work
The scope of the project was to develop a reliable GPS-based computer guidance
system that could be loaded into a panel computer and mounted in cabin of vehicles
operating in open-pit mines. The system developed, automatically warns the driver of
the dumpsite’s edge and the proximity of a nearby vehicle.
As such, the software system, called “VirtualMine”, is written in Visual Basic™,
integrating three different technologies: the Global Positioning System (GPS), wireless
radio networks, and Virtual Reality Modeling Language (VRML).
VirtualMine uses GPS to track the position of the mine vehicle with respect to other
vehicles and to a virtual safety berm using Differential GPS to achieve sub-meter
accuracy and prevent accidents caused by the vehicle driving too close to the berm
edge.
VirtualMine is capable of generating 3D digital contour maps during vehicle
operation. It uses a wireless IEEE 802.11b TCP/IP radio-network system to transfer
position and mine geometry to other vehicles.
|
Colorado School of Mines
|
3
This system incorporates a graphical user interface based on Virtual Reality
Modeling Language (VRML), a powerful 3D internet-compatible graphical engine.
VirtualMine is designed to generate and update 3D digital contour maps of the
mine site on demand using the GPS coordinates generated by the vehicle as it moves
within the mine. Contour maps may also be generated using information from other
vehicles coming through the TCP/IP 802.11 b radio communication network. Thus, in
this system for example, a dozer can transmit updated mine geometry of the working
area to a remote truck, and the truck can use this new terrain profile to update its own
contour map as a reference for its next dump.
In order to make it compatible with any GPS receiver using the National Marine
Electronics Association (NMEA) code, the system integrates an internal algorithm to
convert geodetic coordinates to Universal Transverse Mercator (UTM) coordinates.
Geodetic to UTM transformation requires a series of corrections and complex
formulae based on a spheroid model of the Earth. This algorithm also provides the
software program with the ability to convert Latitude and Longitude into the UTM
equivalent.
Once the system was developed, it was tested by mounting the hardware and
software onto local vehicles operating at the Colorado School of Mines campus. The
system was checked and debugged and further modified for improvements.
Once the system passed the initial test at the CSM, it was taken to a local quarry to
be further tested under more rugged conditions. The quarry chosen for this test was
the Holnam Portland Cement Quarry, near Colorado Springs, Colorado.
Further, additional tests were carried out at Phelps Dodge’s Morenci Mine, near
Morenci, Arizona, to investigate system effectiveness and operator acceptance.
|
Colorado School of Mines
|
4
System description, summary of work, and previous work related to this project are
presented in Chapter 1 along with a literature review based on GPS systems, pseudo
satellites and wireless communications.
In order to acquire highly accurately GPS measurements, different techniques have
been developed. An explanation of these techniques and expected GPS resolution
are reviewed in Chapter 2.
A detailed description of the GPS system, wireless network system used under this
safety software system is cover in Chapter 3. Description of software development,
from its original 2D concept into its current 3D stage, is also given in Chapter 3.
Results and statistics of the GPS accuracy and precision tests carried out at the
Colorado School of Mines survey field and at Portland Colorado are discussed and
summarized in Chapter 4.
The VirtualMine Graphical User Interface (GUI) and software installation are
described in the form of a user manual in Chapter 5. The user manual divides the
GUI into five sections which are described in detail.
Chapter 6 describes a test carried out at the Morenci Mine in Arizona which
consisted in proving vehicle tracking and on-demand contouring. Also proximity
warning systems and real-time map sharing tests were carried out.
Suitability of the system covering pros and cons of this system as well as
conclusions and recommendations are given in Chapter 7. Potential research and
applications based on this system are also described.
|
Colorado School of Mines
|
5
The Visual Basic code written to develop this system is presented and described
for each VB window used under this program as an Appendix to this dissertation.
1.3 Previous Work
1.3.1 GPS Receiver Technologies.
The project requires a GPS system capable of sub-meter resolution. Research on
GPS technologies shows that Differential GPS and Real Time Kinematics (RTK) GPS
are capable of sub-meter accuracy.
RTK GPS is based on processing the carrier phase signal coming from the GPS
satellite. If this signal is further corrected using a local beacon, the technique is called
“Real-Time Carrier Phase Differential” or simply “RTK Differential”.
Four major producers of RTK - capable GPS, were surveyed: ASHTEC, Trimble,
TOPCOM, and LEICA. Based on their technical specifications, all of these GPS units
handle technology based on Dual-Frequency, Differential GPS, and Carrier/Code
phase readings (RTK).
Currently the main GPS products used in the mining industry are Trimble,
ASHTEC, and LEICA. Thunder Basin Coal in Wyoming for example, uses a Trimble
GPS to improve productivity in its mining operation. As described by Long (1998), the
system used in the mine includes a six-channel, L1, C/A-code Differential GPS
(DGPS) receiver Trimble Placer-4400-GPS receiver, with an antenna mounted on an
external post that rises above the truck's roof. Using a Trimble 900MHz
multidirectional radio, each vehicle's onboard computer sends GPS-based position
information back to a dispatch computer, which corrects the data.
|
Colorado School of Mines
|
6
This computer monitors the location of each vehicle in the fleet. The system
analyzes production statistics. The system then correlates this data in order to route
all vehicles efficiently.
Another technique to achieve sub-meter GPS accuracy is to use differential GPS
based on the Omnistar Differential service, which consists of geostationary satellites
broadcasting differential correction signals to specific zones on Earth. Omnistar has
ten permanent base stations in the U.S. and one in Mexico. These eleven stations
track all GPS satellites and compute corrections. The corrections are sent via wired
networks to a network control center in Houston. At the control center, these
messages are checked and sent to a satellite transponder. This occurs approximately
every 2 seconds. A packet will contain the latest corrections from each of the 11 base
stations.
1.3.2 Satellite Signal Emulators / PSEUDOLITES
Reliable sub-meter GPS measures are based mainly on the number of satellites
available for the tracking progress. Due to the geometry of an open-pit (the wall slope
angle and the deep), a successful link between the mine vehicle and the GPS satellite
is compromised by the probability of having four or more satellites visible in the sky.
As discussed in Stone (1999), a technique to increase satellite constellation is the use
of pseudolites.
Pseudolites are advanced ground-based devices that simulate a GPS satellite.
They can be placed in strategic locations around a mine site to provide additional
satellite signals for receivers. LeMaster (1999) explains the use of a new kind of self
positioning transceiver pseudolites.
Pseudolites present challenges that must be covered before they can become
commercially viable, among this is the fact that they might interfere with the Wide Area
|
Colorado School of Mines
|
7
Augmentation Service (WAAS) currently under development by the U.S. Department
of Transportation.
As described in Walter (1994), WAAS is intended to support civil aviation; it uses
pseudolites, combined with GPS, to allow precision landing approaches to airports.
Accordingly, it may be difficult to obtain a license for a pseudolites to broadcast on a
GPS frequency since it could be a source of interference for aerial navigation.
The following table presents a list of a number of companies specialized in
manufacturing pseudolites:
Table 1: List of companies manufacturing pseudolites
BFGoodrich Aerospace New Century, KS
CAST, Inc Billerica , MA
Global Simulation Systems Inc Fort Worth, TX
Gnostech Inc Warminster, PA
GPS Networking, Inc Pueblo, CO
IntegriNautics Menlo Park, CA
L-3 Interstate Electronics Corporation Anaheim, CA
N A VS VS Corporation Colorado Springs, CO
Perkin Elmer Salem, MA
Rockwell Collins Government Systems Cedar Rapids, IA
1.3.3 Wireless Communications
A critical component of this project is the ability to transfer data, or communicate,
between the units being tracked by the satellite and a central office, within a wireless
communication network.
A network of trucks using 900 MHz radios is analogous to a standard network of
computers, but instead of using network cables or modems hooked into the PC, this
system uses IP-addressable Trimcomm 900 radios as the network card or modem,
where the radio modems, besides having the task of handling mining related data
|
Colorado School of Mines
|
8
through a network, also handle GPS differential corrections needed for high precision
accuracy. Currently, these Trimcomm radio modems must be configured to create a
local network by upgrading the radios into a TCP/IP protocol. As described in
Goddard (1998), the radios from Trimble use propriety protocols to make them TCP/IP
compatible.
Wireless data transfer between mobile units is a key factor in mining operations,
either for data transferring or GPS sub-meter technology. Wireless Local Area
Networks (LAN) radios are becoming more reliable and less expensive, and thus a
significant alternative technology to be used in mining for wireless communications
between vehicles and a central base, Rysavy (1999).
Currently there are many companies that provide wireless communication
equipment such as wireless network access-points for the servers in the central office
and PC-cards for the computers loaded on the mobile equipment.
In radio communications, there are three interrelated factors: power, range, and
frequency. Power defines the range of the system and frequency defines the capacity
of data transmission. The power and frequency of the system are regulated by the
government, which does not allow higher power and frequencies due to potential
interference with primary users (Police, Army, etc.), as described in Flood (1999).
Most wireless LANs today use spread-spectrum technology since it allows reliable
communication at much lower signal to noise ratios. Spread spectrum is also used to
avoid interference with other bands: This is the result of FCC rules that allow for
unlicensed operation in a number of radio bands, including 902 to 928 MHz, 2.400 to
2.483 GHz and 5.725 to 5.85 GHz.
|
Colorado School of Mines
|
9
In spread-spectrum wireless radios, the power of transmission is restricted to 1watt.
One needs to obtain an FCC license if more than 1 watt of power is to be used.
Interference with other bands has not been an issue since this spectrum appears as
noise to all but previously fixed receivers. Rysavy (1999) discussed spread spectrum
in detail.
The main challenge in a project like this one is the ability to communicate within a
long distance range between the vehicles and the capacity to transmit large packets
of data. To comply with FCC regulations, this is only possible with a maximum power
source of 1 watt and a low MHz rate of either 900 MHz or 2.4 GHz.
There is a tradeoff relationship between range and data transmission bandwidth
which is related to the frequency and power of the signal; higher frequencies consist
of “dense” energy signals which require more power than “light” energy signals. These
denser 2.4GHz waves require more energy to be broadcasted than the “lighter” 900
MHz waves. Thus, a 900 MHz radio using a 1 watt amplifier has a better range than^a
2.4 GHz radio using the same power.
On the other hand, higher frequencies allow greater rates of data transmission than
lower frequencies, for example a 2.4 GHz radio can transmit 11 Mbps in comparison
with a 900 MHz radio with a capacity of 128 Kbps
1.3.4 IEEE 802.11b
As described by Molta (1999), 802.11b is an extension of ethernet to wireless
communication; it is primarily used for TCP/IP but can also handle other forms of
networking traffic, such as PC file sharing standards.
The IEEE 802.11b specification allows for the wireless transmission of
approximately 11 Mbps of raw data at distances of several hundred meters using the
|
Colorado School of Mines
|
10
2.4 GHz band. (The distance depends on impediments, materials, and line of sight.)
On the other hand, range distance can be increased using 1 watt amplifiers.
IEEE 802.11b started to appear in commercial form in mid-1999, with the advent of
Apple AirPort components manufactured in conjunction with Lucent's WaveLAN
division. (The division changed its named to Orinoco). Eventually, several other
companies e.g., Teletronics, started to build radio cards, access-points and amplifiers
based on this technology. In this dissertation, Teletronics and Orinoco radios were
used.
The client hardware is typically a PC card although USB and other forms of
802.11b radios are also being introduced. Adapters for PDAs, such as Palm OS and
PocketPC based devices, are due out mid-2001.
Each radio may act either as a hub or for computer-to-computer transmission, but it
is much more common that a wireless local area network (WLAN) installation use an
access point dedicated stand-alone hardware with a more powerful antenna.
Several new, compatible protocols are in the process of being released, including
802.11a (54 Mbps over the 5 GHz band), 802.11g (22 Mbps over 2.4 GHz), and
Texas Instruments' PBCC 22 Mbps standard.
The 802.11b radios are considered on this project since they are user-friendly,
simple to deploy and straightforward to configure. 802.11b represents a relatively
inexpensive solution for wireless communications. However, shortcomings in the
802.11b system seem to be a lack of robustness and power, which could impact on
effectiveness and range. As described later in this dissertation, the power/range
constraints present on this radio system has been solved using 1 watt amplifiers and
omni-directional antennas.
|
Colorado School of Mines
|
11
1.3.5 GPS and Wireless Network Based Systems.
GPS systems focusing on mining applications are already being installed in several
mining operations around the world. Some examples are the GPS-based Computer-
Aided Earthmoving System (CAES) developed by Caterpillar, and the GPS-based
dispatching system IntelliMine™ from ModularMining.
As described by Greene (2000), CAES uses on-board computers, software,
centimeter accuracy, real-time kinematics global positioning systems (RTK GPS),
data radios, and receivers. The CAES system consists of two kinds of software:
CAESoffice and CAESon-board.
CAESoffice allows staff to monitor progress in the mine from a central office and
CAESon-board is the software on-board the machine which enables the operator to
view the machine’s location and its work plan on a ruggedized display computer.
CAES in a dozer displays elevation files containing design limits and elevation which
can be seen in plan, or profile views.
METS-manager is the computer interface for translating plans in DXF (AutoCAD) or
American Standard Code for Information Interchange (ASCII) format into CAES
format. Flat and incline elevation designs are a request of the office software or
generated from a digital terrain model (DTM). The designs show operators limits at
working areas and how much they need to cut or fill.
Mining and Earthmoving Technology Systems (METS) manager runs on a PC Windows
platform. Using a data radio system (TRIMCOMM 900 MHz) developed by Trimble,
METS manager transmits the terrain information to the appropriate CAES-equipped
machine. CAES is used primarily to control floor levels and display 2D spatial
information to the drivers.
|
Colorado School of Mines
|
13
As described in ModularMining (2001), the IntelliMine mine management system
provides another tool in mine control technology that optimizes the performance of
mining operations. Used in open-pit mines, IntelliMine is focused in productivity by
integrating haulage fleet assignments using RTK GPS units from ASHTEC, and
spread-spectrum radio communications network called IntelliCom.
IntelliMine has two subsystems related to earthmoving tasks; ProVision-
DozerSystem and ProVision-Excavator. The Dozer and Excavator mobile computer
system are based on GPS technology that includes:
• An Ashtech GG24 GPS receiver
• A 9600-b/s data radio or spread spectrum radio
• An A29K processor
• A GPS/GLONASS choke-ring antenna
• A 10.4-inch computer touch screen.
• A bi-axis inclinometer
• External Ashtech GPS receiver (Excavator System)
IntelliCom is a radio network based on direct sequence spread spectrum (DSSS)
technology at 2.4 GHz, which requires a set of repeaters present in the pit to increase
range.
Provision systems take the output information from a mine-planning package as its
input. The information is downloaded as a DXF file and the DXF file is loaded into the
IntelliMine system. The DXF file contains the target surface, consisting of vectors
representing the desired terrain geometry. This digital geometry of the terrain is called
Triangulated Irregular Network (TIN).
|
Colorado School of Mines
|
14
During the shift, whenever a dozer requests a relief map, the appropriate map
information is sent to the dozer via a wireless radio network. The relief map includes a
dozer 3D-icon, which moves as the dozer moves thereby showing the operator’s real
time position at all times.
As the dozer begins working on the project region, advancing both horizontally and
vertically, dozer coordinate information (x,y,z) is determined. The z coordinates, (the
actual surface elevation) are then compared with the target elevation specified in mine
planning.
The system then color-codes the relief map on the Color Graphic Console (CGC),
showing the operator which areas to cut and fill in order to achieve the target surface
elevation for the region. Areas of the project region that are at the target elevation
appear in green on the relief map, those above the target elevation are red, and those
below it are blue.
The operator cuts the red areas and fills the blue areas, continuing until the entire
project region is green (or at the target surface elevation). The CGC also has a real
time, color-coded elevation gauge that the operator can use as a reference while
working in a region. The gauge shows how far above or below the target elevation the
dozer is currently working.
|
Colorado School of Mines
|
16
According to the NIOSH report, both RFID and radar technology were capable of
detecting obstacles in the blind spots of mining equipment. These show great promise
as an effective technology for a collision warning system.
In cooperation with NIOSH, Trimble is developing a proximity warning system
based on GPS; they have written a two-dimensional interface based on Windows-CE
which is shown in Figure 3. The interface, gives the operator an abstract
representation of his position with respect to a remote vehicle within a safe range.
This is represented as a two-dimensional circle.
[Proximity Warning: RW HAUL TRUCK 12
Ih
^Tn
Figure 3: Photo of the interface taken during the NIOSH proximity warning test.
1.4 Original Contribution
The literature review indicates that there has not been a system developed using
GPS and wireless communication system focused on proximity warning to improve
safety of open-pit mining operations that uses a system to display simultaneously the
local and remote vehicle in the on-board computer. In addition, the review indicated
that systems currently used do not give the operator a 3-D real-time interface, relative
to a mine map as shown in Figure 4.
|
Colorado School of Mines
|
17
NIOSH Vli LudlMine
VirtualMme - Colorado School ot Mines
Updated Updated
Goto T2
r Low (S Med . y
Start Tracking
Lon (X| ; Longtude
Latitude
At(Z| j Altitude ISO View
Plane Did
no tracking
Topo jC\vmgp$<goU
Start Update
DTM VRML Mapt GPS Terminal DaT tr au Bc ak se Vie Pw la nS eafe E&it
SST'
Figure 4: VirtualMine system interface used with mine vehicles
This dissertation is based on “improving safety of off-highway trucks through GPS”
project, which is being developed at the CSM, Mining Engineering Department, in
response to the NIOSH request to improve safety conditions during dumping and
routine haul road operations.
This dissertation offers four major contributions:
• This dissertation follows a unique approach in developing a computer system to
improve safety in open-pit mines based on GPS, wireless networks, and 3D
graphics. The previously developed GPS systems have so far focused on
computer-aided excavation systems and dispatching to improve productivity of
mining operations.
e The integration of 3D on-demand mapping based on Virtual Reality Modeling
Language (VRML), GPS, and wireless communication networks to create a user
|
Colorado School of Mines
|
18
friendly, powerful and inexpensive system to monitor vehicle activity and the mine
geometry to improve safety in open-pit mines is new and is a significant
contribution of this dissertation. None of the GPS systems are currently being used
in the mining industry provide real-time relative vehicle positions with respect to
updated mine topography maps on-demand, nor with respect to other vehicles.
• The proximity warning system developed in this dissertation uses a 3-D sphere
bubble concept to represent the safety zone around a given trench. The
adjustment of the safety sphere radius, relative to the operating conditions of the
vehicle, and the dump surface, is unique and has not been implemented before in
any other system based on GPS and wireless networks, Figure 31 and Figure 47
show some examples of the 3D Sphere applicability during tests at the CSM
survey field and Morenci. This is further explained in Chapter 5.6.4.
• The software implementation of 802.11b wireless networking protocol in the mining
environment is new and provides the capacity to track the equipped vehicles in the
mine from a central base using the TCP/IP protocol. Advantages of the 802.11b
system over the 900 MHz Trimble system are the transmission rate capacity, the
802.11b framework which is based on an open user-friendly architecture, and in
addition, the 802.11b is a less expensive system compared to 900MHz radio
systems.
|
Colorado School of Mines
|
19
CHAPTER II
2 THE GLOBAL POSITIONING SYSTEM (GPS)
The Global Positioning System is a navigation system which is based on 24
Department of Defense (DoD) satellites and four ground stations.
GPS uses the satellites as reference points to calculate positions very accurately
by a method called trilatération. Trilatération is similar to triangulation except that it
uses the lengths of three or more lines, rather than the angles from three or more
points, to locate the GPS receiver. The GPS receivers pick up these signals from the
GPS satellites. Using information embedded in the signal, the receiver calculates the
distance to the four best-positioned satellites to do the trilatération calculation to
derive the coordinate of the receiver.
Differential GPS and Real Time Kinematics (RTK) GPS are two methods used to
augment the calculation of the receiver’s coordinate. These GPS forms are used to
mitigate GPS errors which directly impact on the level of accuracy expected.
2.1 GPS Position: Elementary Explanation
The trilatération calculation requires the distance of the satellites to the receiver in
order to solve for the receiver position (x,y,z). The receiver position is calculated using
at least three equations. Since the coordinates of one satellite are used to linearize
the system, at least four satellites are needed to solve for the trilatération system of
equations. The following example adapted from Hereman (1995) and Navidi (1998)
illustrates the trilatération approach.
Statement of the problem:
The goal is to find the unknown 3D-position of a given GPS receiver, given the
distances from the receiver to the GPS satellites. The (x, y, z) coordinates represent
|
Colorado School of Mines
|
24
centimeters (one foot) of error in measuring the distance to that satellite. For this
reason, the satellites are equipped with very accurate atomic clocks. Even these
clocks, however, accumulate an error of 1 billionth of a second every three hours. To
resolve the satellite clock drifts, they are continuously monitored by ground stations
and compared with the master control clock systems that are combinations of more
than 10 very accurate atomic clocks. The errors and drifts of the satellite clocks are
calculated and included in the messages that are transmitted by the satellites. In
computing the distance to the satellites, GPS receivers subtract satellite clock errors
from the reported transmit time to arrive at the true signal travel time.
Even with the best efforts of the control centers in monitoring the behavior of each
satellite clock, their errors cannot be precisely determined. Any remaining satellite
clock errors accumulate typically to a few nanoseconds; causing a distance error of
about one meter, see Figure 6.
Similarly, any error in the receiver clock causes inaccuracy in distance
measurements. Assume that at a given time the receiver clock has an error of one
millisecond, causing a distance error of about 300,000 meters. If the distances to all
satellites are measured exactly at the same time, then they are all off by the same
amount of 300,000 meters. One can, therefore, include the receiver clock error as one
of the unknowns that one must solve for. In determining the location of the GPS
receiver one has three unknowns (x,y,z) for the position. There are four unknowns:
three components of position and the new unknown of receiver clock error. To solve
for these four unknowns, one needs four equations. Measuring distances to four
satellites provides four equations, (x,y,zj and time, which is used to adjust the clocks in
the GPS receivers to acquire an accurate measure.
Note that the concept of receiver clock being one of the unknowns is valid only if
measurements are taken to all satellites exactly at the same time. By making
|
Colorado School of Mines
|
25
simultaneous measurements to four satellites, it is possible to find the error in the
receiver clock with very good accuracy. A typical clock has a drift of about 1000
nanoseconds every second, but it can now adjust the receiver time to the accuracy of
the GPS clock. Receivers correct their clock every second and keep a corrected time
signal. Four is the minimum number of satellites needed in order to compute position
and time.
Figure 6: Error produced by satellite clock, based on Trimble (1999)
2.2.2 Satellite Orbit Error
The accuracy of the computed position also depends on how accurately the
locations of the satellites (the points of references) are known. The orbits of satellites
are monitored continuously from several monitoring stations around the earth and
their predicted orbital information is transmitted to the satellites, which in turn transmit
to the receivers. The history of GPS shows that the accuracy of the orbital prediction
is on the order of a few meters. This will create a few meters of error in computing the
position.
|
Colorado School of Mines
|
26
2.2.3 Atmospheric Errors
2.2.3.1 Ionosphere
In computing distances to satellites, one measures the time it takes for the satellite
signal to reach the receiver and then multiplies this by the speed of light. The problem
is that the speed of light varies due to atmospheric conditions. The upper layer of the
atmosphere, called the ionosphere, contains charged particles that slow down the
code and speed up the carrier.
The magnitude of the ionosphere’s effect is much greater during the day than
during the night. The magnitude also has a cyclical period of 11 years that reaches a
maximum and a minimum. For the current cycle, the ionosphere has reached its peak
magnitude in 1998 and will reach its minimum in 2004. The cycle will then be
repeated. The effects of the ionosphere, if not mitigated, can introduce measurement
errors greater than 10 meters.
Some receivers use a mathematical model to consider the effects of the
ionosphere. Knowing the density of the charged particles in the ionosphere (as
broadcast by satellites), it is possible to reduce the effect of the ionosphere by about
50%. The remaining error is still significant (5 meters).
The impact of the ionosphere on electronic signals depends on the frequency of the
signal. The higher the frequency, the less is the impact. So if one transmits patterns
simultaneously via two different frequencies, the ionosphere may delay the code on
one frequency by 5 meters and on the other frequency by 6 meters. One cannot
measure the magnitude of these delays, but their difference can be measured by
observing the difference in their arrival time, which in this case translates into 1 meter
of effective distance between them. By measuring this difference and using the known
formula for the frequency dependency of the ionosphere delay, ionosphere effect can
be removed.
|
Colorado School of Mines
|
27
It is exactly for this reason that all GPS satellites transmit information in two
frequencies, known as L1 and L2. Precision receivers track both signals in order to
eliminate the effect of the ionosphere. All non-precision receivers track only the L1
signal. This is one of the main distinguishing features between different types of
receivers. The L1 receivers are also called single frequency receivers, while the
receivers that track L1 and L2 are called dual frequency receivers. Dual frequency
receivers practically remove the ionosphere effects. Figure 7 shows the schematic
representation of possible GPS errors due to atmospheric conditions.
Since the L2 signal is not entirely available to the general public, sophisticated
techniques have been implemented in receivers to extract the code and carrier
information, even with the partial availability of the L2 signal. These techniques fully
satisfy the requirements of the users for non-military applications, while not
compromising the policy and security objectives of the US Department of Defense
(DoD).
28,000
,
200
50
Figure 7: Error produced by atmospheric conditions, based on Trimble (1999)
|
Colorado School of Mines
|
28
2.2.3.2 Troposphere
The lower level of the atmosphere, which contains water vapors, is called the
troposphere. The troposphere slows down both code and carrier signals. The effects
of the troposphere cannot be removed using dual frequency systems. The only way to
remove these effects is by measuring water vapor content, temperature, and
pressure, and then applying a mathematical model that can compute the delay, see
Figure 8.
Clouds
28,000
200,
Figure 8: Error produced by atmospheric conditions, based on Trimble (1999)
2.2.4 Multipath Errors
In measuring the distance to each satellite, one assumes that the satellite signal
travels directly from the satellite to the antenna of the receiver. But in addition to the
direct signal there are reflected signals from the ground and from objects near the
antenna that also reach the antenna through indirect paths, thereby interfering with
the direct signal. The compound signal creates an uncertainty about the true signal
arrival time. If the indirect path is considerably longer than the direct path (more than
10 meters) such that the two patterns of signals can be separated, then the multipath
effect can be substantially reduced by signal processing techniques, see Figure 9.
|
Colorado School of Mines
|
30
measurement. Different satellite geometries can magnify or reduce the errors
described above. A greater angle between the satellites provides a better
measurement. A higher DOP indicates poor satellite geometry and an inferior
measurement configuration.
DOP is a factor inversely proportional to the volume of a body which is formed by
points obtained from the intersection of a unit sphere with the vectors pointing from
the observing site (the vehicle) to the satellites, as seen in Figure 10.
Figure 10: Volume body formed by 4 satellites with respect to the vehicle
on Earth which is used to calculate the DOP, Based on Trimble (1999)
Some GPS receivers analyze the positions of the satellites available based upon
the almanac, and choose those satellites with the best geometry in order to render the
DOP as low as possible. Another important GPS receiver feature is the ability to
ignore or eliminate GPS readings with DOP values that exceed user-defined limits.
Other GPS receivers may have the ability to use all of the satellites in view, thus
minimizing the DOP as much as possible.
|
Colorado School of Mines
|
32
In this case, the weight matrix has been assumed to be a unit matrix. The cofactor
matrix Q is a matrix where three components are contributed by the site position
(x,y,z). Denoting the elements of the cofactor matrix as
qxx qxy qxz
8= goy w 0%
(13)
qxz qyz qzz
The diagonal elements are used for the following position DOP definition:
DOP = -yjqxx + qyy + qzz (14)
2.2.7 Selective Availability
In the last decade the US Department of Defense (DoD) introduced intentional
errors in order to degrade the position accuracy of GPS to about 100 meters. This
intentional degradation was called Selective Availability (SA) and was implemented by
shifting the satellite clocks and reporting the orbit of the satellites inaccurately. SA was
turned off in 1999 and until now has not been reactivated.
2.3 Differential GPS
To achieve sub-meter accuracy, differential correction is needed. Differential
correction requires two sources of data. One source is the raw uncorrected GPS
code-phase data captured by the GPS receiver in the vehicle. The other source is the
corrected GPS information from a base station. The raw GPS data used in DGPS is
based on reading the code phase signal from the satellite. Code phase readings
without differential correction give average accuracy of 50 meters.
|
Colorado School of Mines
|
33
A base station is a stationary GPS receiver in which the position of the antenna has
been surveyed to millimeter accuracy. Since the base station already has an accurate
location, the raw data is corrected for any variances by simply subtracting the raw
location by the known location of the base station. This process is completed on a
second-by-second basis. The correction factor is then broadcast via radio to the
vehicles or rovers equipped with differential GPS receivers, see Figure 12.
Different ial GPS
Radio Signal Correction
Rover
Base known Posil
Figure 12: Representation of Differential GPS (DGPS) positioning. Image composed
using Microsoft ClipArt.
The accurate knowledge of the position of the base directly impacts the accuracy of
the position computed by the rover. If a position is entered for the base receiver that is
off in some direction, then all range errors computed and transmitted by the base
receiver will be off in such a way that the computed rover position will be off by the
same amount and in the same direction as the base.
|
Colorado School of Mines
|
34
As discussed in Parkinson (1996), the distance between the base and the rover
receivers is called the "baseline”. When the baseline is in a range within 10 Km, the
range errors for the two receivers are nearly identical; therefore, the errors calculated
by the base can be used to correct the vehicle position. As the baseline gets longer,
the correlation between range errors becomes weaker. As a rule, it can be expected
an additional one millimeter of error or uncertainty for every kilometer of baseline
when dual frequency receivers are used. For single frequency receivers, this error
increases to 2 millimeters.
The trilatération calculation given in this chapter assumes there are no errors on
the distances (ri,r ,r ,..,rn.) calculated by the GPS receiver. These distances are
2 3
calculated by multiplying the speed of light by the time it took the signal to reach the
receiver from the satellite. As a result, the distance can be miscalculated if the time
signal reported to reach the receiver from the satellite is affected by atmospheric
conditions, clock errors, multipath, etc. These errors can be corrected by differential
GPS.
(X,y2, Z2)
§2 2
A vr zrs)
Ai Xr>
A z
is not shown
Figure 13: Schematic representation of the problem for differential correction
|
Colorado School of Mines
|
35
Adapted from Hofmann (2001) and Parkinson (1996) the following is a simplified
description of the differential correction process where the differential bias (AX# A I#
AZb) is considering the total bias including satellite and receiver error factors. As seen
in Figure 13, represents the position of the satellites, V(x,y,z) is the receiver
position as calculated from the satellites, and B represents the position of the base
station as determined from a given GPS receiver positioned at the base station.
VfxrvyrvZrv) and B(xrByrBZrz) denote the real position of the vehicle and the base station
respectively.
Let us call (xv,yvzv) as the calculated coordinates of the vehicle (V) using the
trilatération technique, and (xB>yBr zB) as the calculated coordinates of the base station
(B) also using trilatération. The exact position of the base station (B), is known from a
precise millimeter triangulation survey: B(xrB,>B,zrs).
The receiver at the base computes the differential error for each vector. AX# AT#
and AZg
AXg = (xra)-W (15)
a 7s = OrsXys) (16)
AZb = (zrB)-(zB) (17)
The differential Correction procedure assumes similar differential errors to occur at
the base station and at the vehicle (V) in relation to the distance between them (the
baseline). There is 1 millimeter of error for each kilometer of baseline.
AXb — AXV (18)
A 7b = A7V (19)
AZ# = AZV (20)
|
Colorado School of Mines
|
37
2.4 Real Time Kinematics (RTK) GPS
Real Time Kinematics (RTK) is a method of correcting GPS readings using carrier
phase measurements. In comparison to Differential-GPS which uses code phase
measurements, RTK uses carrier phase measurements. Correcting carrier phase
measurements can be performed in real-time using a base station or can be corrected
using a post-process method. The process of correcting in real-time carrier phase is
called Real-Time Kinematics (RTK), Allison (1998).
Thus, RTK determines positions by measuring distances to the satellites using the
embedded carrier phase signal of the satellite and then correcting the position in real
time using a base station. Since the carrier phase signal is a millimeter in length, the
expected accuracy of a RTK system is on the order of a centimeter.
The key issue in RTK is based on integer estimation, which consists of finding the
meter mark on the signal. Estimating the integer numbers incorrectly is like reading
the wrong number on the meter-mark. It is like measuring something to be 3.55 meter
when is actually 4.55 meter. The system can read the millimeter-marks very
accurately but misread the meter-mark integer.
After the receiver resolves the integer number correctly, the accuracy of each
position computation is between 0.5 to 2 cm horizontal and 1 to 3 cm vertical
(depending on the antenna multipath rejection capability). All RTK accuracy
specifications from manufacturers are within this range and are based on the
assumption that the integers are estimated correctly.
Resolving the integers correctly is the key in RTK. The issue is how long it will take
to resolve the integers reliably after satellites are locked. If the signals are resolved
|
Colorado School of Mines
|
38
incorrectly, it is like reading the meter-number wrong but continuing to concentrate on
reading the millimeter marks.
The time required to solve for the ambiguities correctly and with a good assurance
is approximately as follows: For short baselines (less than 10 kilometer) the time
required to wait depends upon:
• The level of assurance (or confidence level) that the system requires for correct
integer estimation.
• The number of satellites.
• Whether the system has a single or a dual frequency receiver.
• The strength of the multipath signal (reflection coefficient of ground).
• Multipath mitigation characteristic of the antenna.
The sensitivity of time to resolve the ambiguities of the above factors are described
below quantitatively, in order to understand the relative importance of each factor.
Javad (1998).
• Confidence Level. It may require 10 seconds to resolve ambiguities with 99%
confidence but 100 seconds to get to 99.9% confidence.
• Number of Satellites. It may require 1 second to resolve ambiguities with 15
satellites but 100 seconds with 8 satellites. It may take 10 minutes if there are
only 6 satellites.
e Dual Frequency Receiver. It may require 10 seconds to resolve ambiguities
using dual frequency receivers but 2 minutes with single frequency receivers.
Having dual frequency for short baselines is like having 50% more satellites.
|
Colorado School of Mines
|
Figure 17: GPS receiver and Trimble radio being installed in a mobile truck.
An Xplore PC screen system is used as the on-board computer and display device.
The Xplore computer unit shown in Figure 16 was selected because it is a very
rugged PC-embedded touch-screen that allows the operator to perform software
commands without using a keyboard.
Xplore is a Pentium III PC computer based on Windows 98 that runs at 500 MHz. It
can be mounted in any vehicle and has one serial port and 4 USB ports. It also has an
internal PC card slot which can be used to install the IEEE 802.11b wireless radio
card.
The VirtualMine software is loaded into the Xplore unit in a given truck or dozer.
This software has been successfully tested running in Windows98, Windows Me, and
Windows 2000 operating systems. It requires at least a Pentium II processor running
at 200 MHz.
3.2 Wireless Network System
The wireless network system is TCP/IP compatible and it can use either the
upgraded Trimcomm 900 radios from Trimble or the IEEE 802.11b-compatible radios
from Orinoco.
|
Colorado School of Mines
|
44
The wireless network is used by the system to broadcast coordinates between
mine vehicles, keeping track of their positions in real time. These coordinates are also
used by the system to generate 3D on-demand Digital Terrain Models (DTMs).
Testing is being done to transmit operative data, such as ore/waste, tonnage grade,
truck condition, road condition, etc.
When the system is configured to use the 802.11b radio system, see Figure 18, the
vehicle sends data to an internal network through a TCP/IP-compatible access-point,
the rover (vehicle) uses an 802.11b radio-card connected to a one watt amplifier fed
by 110 volts (AC) power supply, as seen in Figure 18. Under this configuration the
vehicle can be monitored from a central office through an internal network, using the
access-point and the VirtualMine software. In this configuration, a USB adaptor is
used to emulate a PC card-slot to connect the TCP radio card to the computer.
GPS Configuration Using 802.11b System
ROVER Configuration
VirtualMine Software
Xplore PC-Screen
Antenna
GPS Receiver
GPS Antenna
BASE Configuration
oooo
[HE.
O
Access Point
AC
Network
Antenna
Figure 18: Schematic representation of hardware configuration for rover and access-point.
|
Colorado School of Mines
|
48
Trimble system cost is approximately $7,500. The TCP/IP access point used in the
802.11 b system costs between $400 and $2,000 depending of the number of network
nodes capacity, compared to $100,000 (plus) cost for the TCP/IP access point used
under Trimble systems, (based on 2000 prices from Trimble and Teletronics).
3.3 VirtualMine Software System
Development of the VirtualMine software and its visual graphical interface was
carried out using Visual Basic language. The software performs several functions:
• Reading and extracting NMEA code that comes from the GPS unit through a
serial port.
• Converting NMEA code from its original geodetic coordinates (latitude and
longitude) into UTM coordinates.
• Transformation of DXF mine maps into 3D VRML maps.
• Visualization of 3D mine maps.
• Position tracking of local and remote vehicles in real time.
• Generation of 3D mine maps on demand.
• Handling of the TCP/IP wireless communication protocol to transmit and
receive positional and terrain data from remote vehicles.
3.3.1 NMEA code processing
In order to monitor and display vehicle position in a system like this one, it is
necessary to “read” the actual coordinates of the vehicle’s GPS unit to display the
truck position related to a map. These coordinates are read from the GPS unit that is
connected to the system’s computer through a serial port. The system developed in
this dissertation was coded to read those coordinates using the most standard format
used by GPS units, the National Marine Electronics Association (NMEA) format.
|
Colorado School of Mines
|
49
GPS receivers commonly use NMEA protocol, the standard communication format
used in GPS. Under the NMEA-0183 standard, all characters used are printable ASCII
text (plus carriage return and line feed). NMEA-0183 data is sent at 4800 baud.
NMEA data is transmitted in the form of ASCII sentences. Each sentence starts
with a a two letter talker ID’, a three letter sentence ID', followed by a number of
data fields separated by commas, and terminated by an optional checksum and a
carriage return/line feed. A sentence may contain up to 82 characters including the '$'
and CR/LF. See Figure 22.
Since some fields are variable width, or may be omitted, the receiver should locate
the desired data fields by counting commas rather than by looking at character
position within the sentence.
Once native NMEA-code is received on the PC port, a VirutalMine subroutine
reads, extracts, and processes the GPS NMEA coordinate data to display vehicle
position.
The process of reading and extracting the NMEA code containing longitude,
latitude, and altitude can be monitored using the interface shown in Figure 22. This
process is carried out dynamically as the data comes from the GPS receiver that is
connected to the computer’s serial port.
A more detailed description of this process, and references on how to convert
geodetic coordinates into UTM coordinates, are given in detail in 5.6.2, under the
VirtualMine Graphical User Interface (GUI) in Chapter 5.
|
Colorado School of Mines
|
51
Figure 23 shows the dump point, defined at a certain distance with respect to a
safety line that can be seen as a “digital” safety berm to the dump’s edge.
Secfron View
*
Figure 23: Cross-section of the dumpsite showing the relative position of the truck
represented as a dot against a buffer safety zone (2D interface).
The software code receives the position of the truck from the GPS system and
displays it on a previously uploaded DXF topo map of the dumpsite. The location of
the safety berm is also displayed with respect to the edge of the dump.
The 2D system is designed such that two separate views of the dumpsite are
generated in the graphical interface: the plan view on the top part of the screen and
the equivalent section view just beneath (see Figure 24). Using this format, the driver
can track, at the same time, both his vertical position with respect to the dumpsite and
his horizontal position. The program also displays a virtual line representing the safety
boundary of the truck with respect to the dumpsite. The line distance with respect to
the edge of the dumpsite varies according to the truckload characteristics and also
with respect to the soil conditions.
Both the mine geometry and the dump point information can be fed into the system
by a radio modem link.
|
Colorado School of Mines
|
54
3.3.3 VirtualMine System Software Development using 3D Graphical Interface.
VRML is an open standard for 3D multimedia to share virtual objects on the
Internet, Andrea (1997). As the second stage of software development for this project,
a new 3D concept was incorporated into the system. For this, a new software code
was developed in Visual Basic incorporating The Virtual Reality Modeling Language
(VRML) software library.
Before its official standardization, VRML became the standard for sharing and
publishing data between CAD, animation, and 3D modeling programs; virtually every
one of those programs now exports VRML or has a utility or plug-in to convert its
native file format to VRML. VRML is included or referenced in the upcoming MPEG-4
standard, Java3D, and in other developing standards.
The ability to talk and work in a 3D shared virtual space was one of the earliest
motivations of the VRML pioneers. The approach followed to construct the 3D
graphical interface for this project was to first create a Visual Basic program that could
translate DXF files (contour lines in 3D poly-lines format) into VRML 3D lines.
Thus, a mine map defined in a DXF format can now be visualized in a 3D view,
which can be dynamically viewed in real time, instead of using 2D sections.
|
Colorado School of Mines
|
63
CHAPTER IV
4 GPS ACCURACY TESTS AT CSM SURVEY FIELD
In order to assess the quality of the GPS results, accuracy and precision tests were
performed at the CSM survey field.
Accuracy tests focused on measuring the departure of GPS measurements from
the true value. Precision tests on the other hand, measured the repeatability of the
data. The difference between accuracy and precision is known as "bias" or
"systematic error". For example, taking large amounts of data will improve the
precision of a sample mean but will not remove its systematic error.
4.1 Accuracy Tests
The coordinates calculated by the GPS receivers were compared to a known
surveyed point at CSM survey field known as the Gaby point. The UTM-WGS84
coordinates of this point are: Easting 481082.353, Northing 4398932.119, and
Elevation 1811.83248.. Equivalent Latitude is 39° 44’ 24.27265” N, and Longitude is
105° 13’ 14.83589” W.
Two GPS receivers were used during the testing: the Trimble-4400 and the
Trimble-AG132. The 4400 receiver measured results based on reading satellite carrier
phase signals L1/L2. The AG 132 receiver measured results based on code phase
signal using differential correction.
The differential correction signal unit was acquired in real time using the differential
correction service from Omnistar. The accuracy expected from Omnistar depends on
the quality of the GPS receiver used; that is, a standard class GPS will give larger
|
Colorado School of Mines
|
64
semi-random results relative to the true position than a "commercial quality" receiver.
The better commercial GPS receivers can achieve horizontal errors of less than a
half-meter for 67 to 73% of the time, less than a meter 95 to 97% of the time and less
than 1.5 meters 99% of the time. Vertical error will be 2 to 2.5 times greater than the
horizontal error, see Omnistar (2001 ). The expected standard deviation of the error
acquired using differential correction has a standard deviation of approximately 0.5
meters.
The Omnistar Differential service consists of geostationary satellites that broadcast
differential correction signals to specific zones on Earth. Omnistar has ten permanent
base stations in the US and one in Mexico. These eleven stations track all GPS
satellites and compute corrections. The corrections are sent to a network control
center located in Houston via wired networks. At the control center, these messages
are checked and sent to a satellite transponder. This occurs approximately every 2
seconds. A packet will contain the latest corrections from each of the 11 base stations,
see Huff (1995).
The accuracy test carried out at the CSM survey field consisted of precisely
positioning the GPS antenna over the “Gaby” point to log and compare the
coordinates generated by the GPS receiver. Figure 35 and Figure 36 are pictures
taken during the Gaby point accuracy tests.
|
Colorado School of Mines
|
66
Accuracy tests using the AG 132 and 4400 GPS receivers were carried out at 10:00
a.m., 4:00 p.m., and 1:00 a.m. to check for GPS accuracy based on different times
and based on different Dilution of Precision (DOP) factors.
Accuracy tests using the AG 132 GPS receiver, that uses differential correction,
under an average DOP of 2.07, based on all collected samples, averaged a bias of
0.304 m. Easting, 0.734 m. Northing and a vector departure of 0.811 m. in relation to
the surveyed Gaby point. Elevation bias using all the samples was 5.46 m. with a
standard deviation of 0.41 m.
Under a DOP of 2.0 present during the test at 10:13 AM, the AG 132 GPS receiver
averaged a precision of 0.40 m. Easting and 0.45 m. Northing in relation to the
surveyed Gaby point. Elevation was biased by 5.75 meters within a variation of 0.1 m.
During the 4:00 p.m. test, with a DOP of 1.9, the AG 132 averaged a bias of 0.3 m.
Easting and 0.89 m. Northing. On the test at 1:00 a.m. with a DOP of 2.3, the AG 132
unit averaged a bias of 0.10 m Easting and 0.17 Northing, see Figure 38 and Figure
39.
The 4400 GPS receiver, which uses L1/L2 carrier phase readings under an
averaged DOP of 2.07, based on all collected samples, averaged a precision of 1.398
m. Easting, 1.714 m. Northing, and a vector departure of 1.92 m. in relation to the
surveyed Gaby point. Elevation bias using all the samples was 32.9 m. with a
standard deviation of 3.85 m.
Under a DOP of 2.0 present during the test at 9:55 AM, the 4400 GPS receiver
averaged a precision of 1.46 m. Easting and 1.33 m. Northing in relation to the
surveyed Gaby point. Elevation was biased by 34 meters from the Gaby point, within
a variation of 0.20 m.
|
Colorado School of Mines
|
71
Table 4: Dilution of Precision (DOP) 24 hours, Nov. 11, 2001
1 GPS—CKPGN I GF3 DILUTION OF PRECISION I 3KG-GI3S
I LATITUDE ; N 39 44 24.000 DATE: 11.11.2001 I
I LONGITUDE : E 105 13 14.000 DAY OF YEAR: 315 I
I EL--HEIGHT: 1340.000 ELEVATION CUT: 10 I
I TIME I I I
CJT) I VISIBLE SATELLITES — BEST GEOMETRY(7) I GDOF PLOP HDCE 1
I 0 0 I 5W 9* 14* 15* 17 18* 21* 23 26- I 2.2 1.9 1.2 I
I 0 30 I 5 9* 14 15- 17* is* 21- 23 29- I 2.7 2.3 1.2 I
I 1 0 I 5- 9* 14 15* 17* 1S- 21» 23 29* 30 I 2.6 2.2 1.2 I
I 1 30 I 5 9- 14 15* IS* 21* 23* 25- 23* 30 I 2.1 1.3 1.0
I 2 0 I 5 9* 11* 14" 18- 21* 25* 29* 30 I 2.0 1.8 1.0 I
I 2 30 I 5* 11" 14» 18* 21- 25* 2 9* 30 I 2.1 1.2 1.1 I
I 3 0 I 5* 11* 14* 21* 25* 29* 30" I 2.3 2.0 1.2 I
I 3 30 I S'* 14* 20» 21* 22- 25* 2 9* 30 I 1.8 1.6 1.1 I
I 4 0 I 6' 14* 20» 21* 22 25* 29» 30" I 2.1 1.8 1.2 I
I 4 30 I 1* 6* 14* 20 22* 25- 2 9* 39- I 2 .2 1.9 1.2 I
I 5 0 I 1* 6 14* 20* 22* 25* 29* 30* I 2.5 2.2 1.2 I
I 5 30 I 1* 3* €» 14 20* 22* 25- 29* I 2 .1 1.3 1.0 I
I 6 0 I 1" 3" 6* 13* 20* 22» 25* 29 I 1.9 1.7 1.0 I
I 6 30 I 1* 3* 13* 20* 22" 25- 29» I 2.6 2.3 1.2 I
I 7 0 I 1* 3* 13* 20* 22" 25* 31» I 2.7 2.4 1.2 I
I 7 30 I 1* 3* 13» 22* 25* 27» 31» I 2.3 2.0 1.2 I
I S 0 I 1* 3* 13 IS 22* 25* 27- 31* I 2.2 2.0 1.1 I
I 8 30 I 1* 3* 13» IS* 22* 27* 31* I 3.0 2.6 1.3 I
I 9 0 I Ie 3" 8* 13* 22* 27* 31» I 2 .2 2.0 1.0 I
I 9 30 I 1" 2" 3 S* 11 13* 22* 27* 31" I 1.8 1.7 1.0 I
H
gIH BH8 H m g ■
a
m ËS
9 EB 033Bfj
i ii 0 i 2 3 11 27 25 31 I I
i ii 30 i 2» 7* S* 27* 2o, 31* I 2.7 2.3 1.2 I
i 12 0 i 2* 7* a* 11* 27* 23* 31» I 2.5 2.2 1.2
i 12 30 i 2* 7- 8» 11* 20* 27» 28* I 3.9 3.3 1.7 I
i 13 0 i 2* 4* 7* 5» 11- 20* 25* I 3.0 2.5 1.5 I
i 13 30 i 2- 4 7» a 9* 11* 29* 25- I 1.9 1.7 1.1 I
4* :
i 14 0 i 2" 7* 9" 11* 20* 28* I 2.9 2.5 1.3
i 14 30 i 2" 4* 7" 9" 20» 24* 28- I 2.6 2.2 1.1 I
i 15 0 i 2" 4* 5» 7* 9» 20- 24 28» I 2.1 1.3 1.0 I
i 15 30 i 2- 4" 5» 7* 20* 24- 23- I 2.0 1.3 1.0 I
0 B H E
BESmQgmH8 sHs HI 1
i 17 0 i i* 4* 3* 7* 10* 13 24- 30* I 2 .3 2.0 1.0 i
i 17 30 i 4- 5* 7* 10* 13* 24* 30" I 2.5 2.2 1.1 i
i 13 0 i 4- 5* 6* 10* 13* 24* 30* I 2.6 2.3 1.4 i
i 13 30 i 4' 5* 6* 10* 13* 23* 24* 30 I 1.9 1.3 1.2 i
i 19 0 4* €* 10* 23- 24* 26- 39* I 2.7 2.4 1.3 i
i 19 30 i 6 10 17 23 24 2€ I i
i 20 0 i 6» a* 10* 17* 23* 24* 26* I 2.9 2.5 1.3 i
i 20 30 i 6 e* 10" IS 17 IS 23- 24» 26" I 2.1 1.9 1.1 i
% 21 0 i 6* 5* 10 IS* 17* IS" 23* 26» I 2.5 2.2 1.4 i
i 21 30 i 6- 9* 10» IS 17 IS 23- 26- I 2.8 2.4 1.2 i
i 22 0 i 6- g. 10» IS 17» IS 21* 23» 26* I 1.9 1.7 1.1 i
i 22 30 i 3* 9- 15- 17* 18* 21* 23 26- I 2.6 2.2 1.5 i
i 23 0 i 3" 9* 14* 15- 17- 18* 21 23 26* I 2.3 2.0 1.3 i
i 23 30 % 9* 14» 15* 17* 18* 21* 23 26* I 2.6 2.2 1.3 i
i 24 0 i 5 9* 14* 15 17* 18* 21" 23 26- I 2.2 1.9 1.1 i
|
Colorado School of Mines
|
72
4.2 Proximity and Precision Tests
Proximity tests consisted of repeatedly checking the vehicle’s GPS position with
respect to a known geometry in the field and its equivalent digital map loaded in the
system.
The geometry used for the test was a regular grid of 6 x 24 meters, defined by
square units of 2 x 2 meters, as shown in Figure 40. The grid was physically marked
at the CSM survey field (see Figure 41) based on two known UTM points.
Point 0, 0: Easting 481094.48 m, Northing 4398935.22 m, and Elevation 1811.18
Point 0, 24: Easting 481083.52 m, Northing 4398957.40 m, and Elevation 1811.18.
00
02
__________________________________________________04
06
24 22 20 18 16 14 12 10 08 06 04 02 00
Figure 40: Grid used for precision test
The computer graphic grid was generated using AutoCAD and then transformed
into the VRML format used by VirtualMine. Once the grid was loaded into the
computer system, the virtual safety berm was positioned at line zero, emulating the
edge of the dumping point, see Figure 42.
|
Colorado School of Mines
|
74
The vehicle was equipped with the AG 132 Differential GPS receiver and positioned
within the grid on repeated occasions during a one-week period to check the
consistency of its reported position with respect to the computer grid. Positions of the
truck as reported by the GPS receiver were logged and photo-captured at different
distances in relation to the line (0,0) in order to be compared to its equivalent real grid
position, as seen in Figure 43 to Figure 46.
The 3D model of the vehicle used in the system was modified to match the
dimensions of the Jeep Cherokee truck used during this test. The dimensions used to
simulate the vehicle within the system were 4.4 x 1.7 x 1.7 meters. Dimensioning the
model of the vehicle to the dimensions of the real vehicle was performed in order to be
consistent with the position of the GPS antenna on both the model and real vehicle,
since in both cases the GPS antenna is located at the center of the truck. The GPS
antenna was mounted on the jeep and positioned at its center (2.2 x 0.85 m) which is
consistent with the position of the antenna in its equivalent 3D model. Note that GPS
coordinates reported in the computer system are based on the position of the GPS
antenna.
The following series illustrations show how the precision test was carried out at the
CSM survey field. The pictures show the position of the vehicle as seen on the
computer screen and at the actual survey field, at interval distances of 8 and 4 meters
to the edge of the dumping point.
Table 5, is the tabulation of results obtained from the precision test. Last column
(Biased) is the offset distance of the truck reported by the GPS receiver with respect
to the real position on the surveyed grid.
|
Colorado School of Mines
|
77
Table 5: Position of vehicle with respect to grid, (Coordinates are based on the 0, 0 grid point).
Test ID DOP GPS GPS Grid GPS Grid Truck Biased
Eastina Northina Truck Position
Nov11-01 1.9 481094.94 4398943.51 7.8, 3.6 8.3, 3.0 0.5, 0.6
Nov11-01 1.9 481096.61 4398939.42 4.2, 3.8 3.8, 3.1 0.4, 0.7
Nov 14-01 2.0 481096.58 4398939.48 8.0, 3.0 8.4, 3.1 0.4, 0.1
Nov 14-01 2.0 481094.81 4398943.21 4.0, 3.0 4.4, 2.7 0.4, 0.3
Nov 15-01 2.3 481096.35 4398938.68 4.0, 3.0 3.9, 3.5 0.1, 0.5
Nov 15-01 2.3 481094.69 4398942.09 8.0, 3.0 7.8, 3.7 0.2, 0.7
Nov 15-01 2.3 481093.19 4398945.59 12.0,3.0 11.5, 3.8 0.5, 0.8
The consistency of the distance between the vehicle and the dumping point, as
reported by the GPS receiver was tested in order to check the system proximity
warning capability in relation to the GPS receiver. The safety berm represented by the
3D plane was positioned at a point coinciding with the point (0, 3) on the grid, as seen
in Figure 47.
Test results show an average accuracy of 2.3 meters in relation to a known point
using the 4400 GPS-receiver (L1/I2 carrier phase). On the other hand the system
achieved sub-meter accuracy with an average error of 0.8 meters when using the
AG 132 Differential GPS-receiver (L1 code phase).
Tests also indicated that precision with respect to a given geometry is consistent.
The system was able to pinpoint the vehicle over the 24x6 grid with sub-meter
accuracy using the AG 132 GPS-receiver and Omnistar differential correction.
In addition, the system was tested for close proximity to a dumping point, using the
grid as a reference and positioning the safety berm at coordinates (0, 3) of the grid.
This test was carried out using a safety sphere radius of 5 meters, see Figure 47.
|
Colorado School of Mines
|
82
5.1 Software Installation
To install VirtualMine on your PC, use the installation CD included in this thesis and
open the CD drive in your explorer and run the setup.exe installation file and follow the
steps.
After installation is complete, install the 3D Virtual Reality Engine, which is located
in the Program Files - VirtualMine folder called: cortvrml.exe. There is also a copy of
this file on the folder called VMGPS.
VirtualMine uses 3D VRML models which are read from a folder called VMGPS
directly under the C:/ directory. To create this folder on your own hard drive under C:/,
copy the VRML folder preloaded on the installation CD onto your hard drive directly
under C:/.
The contouring routine in VirtualMine is called from a Surfer library. Thus if the user
is going to execute contouring the user needs to install Surfer as well.
VirtualMine initially uses a surfer contouring routine to internally generate contour
lines. Then VirtualMine imports these contour lines into the 3D graphic screen using
the 3D VRML format transformation.
5.2 System Hardware Components
Mobile Computer System
• GPS Trimble 4400 L1/L2 GGA-NMEA string
• AG-GPS 132 Trimble L1/L2 RTK GGA-NMEA string
• Trimble GPS-Antennas, L1/L2.
• Omni-directional Antenna, Base 12dBn special sealed
• Omni-directional Antenna, Rover 3dBn
• Xplore Genesis Computer-Screen Pentium III 550 MHz
|
Colorado School of Mines
|
83
• Sony Notebook Computer Pentium III 750 MHz
• USB/Serial adaptors
Wireless Network
• Orinoco PC-Card wireless card 802.11b, 11 mbs, 2.4 GHz
• Amplifier for PC-Wireless Cardl watt
• Wireless Network Access-Point 802.11 b
Software
• WindowsME/2000/XP
• VirtualMine software
• SURFER software
System Estimated Cost (one unit): $10,000
5.3 Section Description of the VirtualMine GUI
The main window is divided into 5 sections:
Section 1 handles VRML maps: loading, visualization, and filtering. It is designed to
allow the user to freely choose any VRML map from any location in the local system
or network.
Section 2 contains controls related to the mine vehicle and is used to load the
VRML model of the vehicle. It also displays and hides the safety truck sphere and
displays the actual distance from the truck to the safety plane and the actual
coordinates of the current truck’s position. It is also used to log coordinates and to
create and visualize Digital Terrain Models (DTMs) of the new surface as they are
generated by the vehicle.
|
Colorado School of Mines
|
86
The Updated Contour and Updated DTM buttons are used to load the contour map
that was previously generated by the vehicle (using the generate contours option
described later) into the 3D graphic window.
5.5 GUI-Section 2
This section is used to load and monitor the position of
the mine vehicle as well as to generate and update 3D
Hide Safe Range
contour maps on-demand. The Add Trk button loads into
r Low Med High
the graphics interface the 3D truck model. Once the truck
Start Tracking is loaded into the interface, the Track Pos. button is
available to start tracking the vehicle’s position.
Lon (X) (Longitude
Lot(Y) jLatitude
The system will now start monitoring truck position
Alt (Z) |Altitude
and distance with respect to the safety boundary. (Note:
Plane Dist.
before tracking vehicle distance to the safety boundary,
Il56 045530100854
the user must load the safety plane using the View Safe
Plane option).
Topo jC:\vmgps\gold
Once the truck is loaded, a default safety sphere
Start Update
(safety bubble) around the truck is also loaded. This
bubble represents the critical distance from the truck to
Generate Generate
Contours DTM
the safety berm plane.
Figure 55: Section 2 VirtualMine
Since this distance is considered under a 3D
environment in any direction from the truck, the concept of a safety sphere is used.
The sphere represents 3 levels of safety according to mine and truck conditions: Low,
Medium, and High. As the user defines and changes the safety parameters and safety
distance, the sphere will change its dimensions
|
Colorado School of Mines
|
87
Sphere dimension is defined by the sphere’s radius, which is the critical safety
distance.
Once the truck, the safety sphere, and the safety plane objects are loaded and
defined, the user can monitor distance and position with respect to these objects.
If the truck’s safety sphere collides with the safety plane, an alarm will automatically
warn the driver that he/she is entering into a dangerous spot in the mine. The
detection of potential collisions of safety spheres can also be used in vehicle proximity
warning.
Section 2 holds the real-time contouring feature of VirtualMine. The Topo
command (Topography) prompts the user to define the location of the file where
survey information will be stored. This file can be a new blank file or can be a file
containing the survey information of a previous surveyed shift.
The file is a comma-separated value (csv) consisting of {x,y,z) values, which
represent the area being contoured. The update process will append the file with the
new {x,y,z) data being generated by the vehicle’s GPS receiver.
The StartUpdate command initiates the process of creating or appending the
survey file. Once the area to be contoured has been covered by the mine vehicle, the
user can stop the process by hitting the same command again.
At this point, a new updated csv file is created and stored, to be read by the next
process (contouring).
|
Colorado School of Mines
|
88
This next process is trigged by executing the Generate Contours command, which
executes the following steps in the background:
First, the ASCII (csv) file containing the vehicle (x,y,z) information is accessed and
interpolated into a regular contour grid using a contouring routine adapted to Visual
Basic from Surfer™ software. This process generates a grid file (grd) defined by
contour lines. The contour lines are exported into a dxf file (dxf). DXF files are ASCII
files containing the information of the contour lines as a series of points (x,y,zj and
defining the end of each contour line.
The DXF file is read and processed by VirtualMine to generate a 3D VRML (wrl)
representation of those contour lines. The VRML is the format used by VirtualMine to
display 3D graphics on the computer interface. This process is described in the
following Table 6:
Table 6: Steps carried out by VirtualMine to generate contour maps in 3D.
Stage 1 2 3 4
File generated extension CSV grd dxf wrl
Process Generation of Regular grid Contour lines Export of contour
survey data generation extraction in DXF lines into VRML
|
Colorado School of Mines
|
89
5.6 GUI-Section 3
o
GPS Truck View Safe
VRML Maps Peer Comm Run Sim
Terminal DataBase Plane
Figure 56: Section 3 of VirtualMine
Located in the lower region of the main window, Section 3 presents the controls
that run several subroutines:
VRML Maps: Generation and definition of 3D maps (VRML format)
GPS Terminal: Reception and processing of data coming from the GPS unit, which
is connected to the PC screen serial port.
Truck DataBase: Vehicle and mine database communication between the rover
and the base.
View Safe Plane: Control and manipulation of the safety boundary berm.
Green Light: Safety vehicle status. This light will turn red and produce a sound
when the vehicle gets too close to the virtual safety berm.
Peer Comm. Since this package is TCP/IP server/client enabled (it can be used as
the rover terminal or the base terminal), it has a Send Position to Base option, which
is used by the rover to establish a TCP connection with the base.
The following is a detailed description of each feature present in Section 3 as seen
in Figure 56.
|
Colorado School of Mines
|
90
5.6.1 VRML Maps: 3D Surface Generation and Transformation Interface
Import Surface _ n|x|
RASTER 2 SOUD DXF 2 VRML
Load | Jc\ LoadDXF C:\
|CÂ Def. VRML Jq\
DTM Surface j |q\
Filter Coords Under
X |-9999999 Y {-9999999 Z {-9999999
Coords Found
DXF Surfer to DXF ACAD 12 to
Generate Solid
VRML VFlMUbetal
Load WRL1 jc;\VMGPS\golden_solid.wrl ViewWRLI
Load WRL2 |c:\VMGPS\golden_contour.wrl View WRL2
Load WRL3 jC:\VMGPS\vrmloutcontlocal.vvrl ViewWRL3
Load WRL4 jC:\VMGPS\DTMJog_truck1 .wrl View WRL4 Exit
Figure 57: VRML Maps Interface
With this option, the user can create 3D maps in VRML format. Special
considerations must be taken to check for the coordinate projection used on these
maps. It is very important to identify which system of projection is being used on the
map you are going to transform. VirtualMine uses Geodetic or UTM coordinates to
pinpoint vehicles in a 3D space, as represented by the VRML map created here. If the
map is based on a different projection system, then the 3D visualization will be
erroneous. The user must convert the map coordinates to the appropriate projection
system (see GPS Terminal Option).
RASTER 2 SOLID creates a VRML 3D grid file from a 3D grid raster file.
|
Colorado School of Mines
|
91
Load: prompts the user to define the location of the INPUT grid file. Note that this
algorithm is written to read SURFER grid files in ASCII format (extension grd).
Def. VRML: prompts the user to define the location of the output VRML file
(extension wrl)
DTM Surface: VirtualMine allows you to create “textured 3D solids”. This is
accomplished by defining an image file in this text box. The image file must
proportionally represent the area of the solid being created.
VirtualMine will superimpose this image over the 3D solid, such that every 2D point
represented in the image will be presented with its equivalent elevation value
according to the grid file.
To accomplish the texturing of the DTM map, the user must be very careful to
check that the image represents exactly the grid file area. Otherwise, VirtualMine will
stretch the image and force it to fit the area, thereby misinterpreting the image with
respect to grid file dimensions.
Another issue is that the image to be used as the solid texture must be previously
mirrored along the X-axis using an imaging program. This is due to the particular way
that 3D images are graphed on the interface of VirtualMine, where the Y-axis runs
perpendicular to the computer screen plane and where the negative values are
increase towards the outside of the screen.
Raster2Solid: Executes the routine.
|
Colorado School of Mines
|
93
5.6.2 GPS Terminal: GPS Signal Control Interface
5 GPS Terminal
File ÇommPort MSComm Call
olU *|j*| s| JEl Hide i Connect
IM 1
Enter Dec Places for Degree and Minutes to be read from GPS data (Latitude and Longitude)
Deg. Dec Places MinNumDigits Deg.Dec. Places MinNumDigits AltNumDig
r i n ¥ I11 r
Longitude GPS Latitude GPS Altitude M. GPS
jDegrees (Minutes (Degrees Minutes iMeters
p Projection System to be used
r Geographic a UTM Longitud Latitud Cent Merid. UTM Zone
vv F Zj [Ô
Ellipsoid name Equatorial Radius Square of Eccentricity
(Ellipsoid "3 Refresh F
Easting Northing Attitude
F F F
—
Figure 58: GPS Terminal Interface
The GPS terminal is used to initiate the GPS / PC data communication through any
of the serial ports present on the PC computer. The terminal reads ASCII NMEA code
coming from the GPS unit. It specifically reads for NMEA / GGA format.
The user must customize the GPS unit to output this type of format; otherwise, the
system algorithm must be modified.
To start the terminal, use the Connect command or simply click on the red light at
the right top of the window.
|
Colorado School of Mines
|
94
Select the appropriate port under the CommPort menu to begin communication.
NMEA GGA code ASCII characters will display on the screen and the system will
automatically extract longitude, latitude, and altitude.
By default the system will read 3 characters from the longitude data, which must
represent the degrees of longitude, and will read 11 characters representing the
minutes of longitude. The same process is carried out to process latitude data. By
default, the altitude value is 8 characters.
Since the system is prepared to use different brands of GPS units, the Output
NMEA ASCII code may vary. Therefore, before transmitting information to the
visualization routine, the user must verify that the proper number of digits is present.
Likewise check for commas or other special characters that should not be present on
the Longitude GPS and Latitude GPS boxes, and proceed accordingly to modify the
number of digits present by default on the first row of text boxes in the GPS window.
Once the data is clean and read into the boxes of longitude, latitude, and altitude
the user can decide either to process it as Geographic coordinates or as Universal
Transverse Mercator UTM projections.
Geographic data will actually go directly into the visualization routine as degrees
and minutes of degrees. Some users prefer this data format in order to be consistent
with their reference maps in geodetic coordinates. However, the user must be aware
that since altitude value is given in linear units (meters) the altitude should be
converted or transformed into an equivalent earth arc in degrees in order to be able to
visualize the object or vehicle in 3 dimensions, giving an inexact 3D visualization.
|
Colorado School of Mines
|
95
UTM projections are the transformations of coordinates of any point positioned in a
spheroid surface into a rectangular 2D plane. The UTM grid splits the world into 60
zones that are 6° wide, (360°) around the equator. This is an accurate method of
representing a spheroid surface into a 2D Plane.
In 1866, Clark developed a model of the Earth on which to base these UTM
projections (known as Clark-1866). USGS maps are based on Clark, however, with
the development of very advanced survey equipment and satellites, new Earth models
or Earth spheroids have been developed and are now being used for GPS mapping.
VirtualMine can transform default geodetic coordinates coming from the GPS unit
into several UTM projections including Clark 1866 and WGS-84 systems. Conversion
formulas were coded in Visual Basic based on formulae available at
http://www.gpsy.com/gpsinfo/geotoutm/. A book on the subject, with many of the
conversion formulas, was written by Snyder (1982).
The user must know which projection system is used in the map. If the mine map is
in a local state coordinate system, it must be transformed into UTMs before using the
VirtualMine VRML maps option discussed earlier.
See the transformation procedure followed by VirutalMine in the following example
where the latitude and longitude of the “Gaby” Point at the CSM survey field are
transformed into its equivalent UTM-WGS84 coordinates:
The coordinate values of the “Gaby” point are:
Latitude: 3? 44’ 24.27265”
Longitude: 105° 13’14.83589”
Altitude: 1,811.81 m.
|
Colorado School of Mines
|
99
5.6.3 Truck Database Interface:
% TRUCK MIN F DATA
Nj 4 Datai ► M Data2 ► M
TRUCKID L0CATI0NX L0CA1* TRUCKID »
► BD1 999999.082 18437F ► BD1
BD10 676203.275 18437:— BD1
BD100 676061.221 184407 BD1
BD101 675997.513 18441 BD1
BD102 676524.936 184321 BD1
BD103 676540.603 18432: BD1
BD104 676390.698 18432: BD1
BD105 67651Z102 184306 BD1
BD106 676071.353 18441: BD1
BD107 676063.655 184421 BD1
BD11 676127.682 18437: BD1
BD110 676599.766 18436 BD1
BD111 676072.469 18441:'" BD1
<1 ►i <1 J ►JZ
Execute SQL O Load DB fc:\
Meet * from Truck_HEADER where LO CAT 10 NX > 676600
Hide
Figure 59: Database interface
This window interface is in beta status. It is prepared for an eventual file-sharing
option under a TCP/IP protocol. The form presents two database spreadsheets linked
to each other, so the user can visualize, in both windows, information related to the
same database.
LoadDB: This command is used to define the server database file across the mine
network, where all data will be stored. Ideally, this file will be centralized at the main
server computer to which all trucks present in the operations send field data.
This window is SQL-compatible, meaning that the user can query using standard
scripting SQL language as shown in Figure 59.
|
Colorado School of Mines
|
100
5.6.4 VIEW SAFE PLANE: Virtual Safety Berm Control Interface
i Safety B... g Q
Vie* Safe Plane
Plane Loaded
X Plane 481000
Y Vert. 4399800
Z Plane |Ï8V3
Y +
Xw Xe
Y-
Z- Z +
Rot- Rot +
Exit
Figure 60: Safety berm
The View Safe Plane command displays this form which contains commands to
load and control the 3D position of a virtual safety berm represented by a 3D plane.
Once the berm is loaded, vehicle position is constantly monitored with respect to
the position of the berm. VirtualMine then calculates the distance between the vehicle
and the centroide of the safety berm in a 3D space considering the (x,y,z) coordinate
values. If the distance between the vehicle and the safety berm is reduced to the
equivalent radius of the safety sphere, the alarm is triggered. This distance is reported
continuously to the main interface.
This process is taken to the graphical interface as the 3D bubble concept. For
example if the bubble crosses the safety berm, the driver is warned of a dangerous
|
Colorado School of Mines
|
102
5.6.5 PEER COMM: Peer to Peer Communication Interface
The PeerComm option of VirtuaMine’s
* Peer A d x l
-
LOCAL PORT USED Control Panel provides all the controls to
11001 r t
maintain a peer-to-peer TCP connection
Remote Unit Remote Port
between two vehicles using a wireless radio
|l 38.67.42.102 |
system.
Commandl
XRec Received X
In order to execute a successful connection,
YRec Received Y
IlMMIinillllllllf the Local port must first be defined as a unique
Z Rec j Received Z
Rot Rec jReceived Rot local port relative to other vehicles in the
X 2 send {Longitude system.
Y 2 send Latitude
Altitude
The remote unit box is used to define the IP
Rot 2 Send |Rotation
number of the remote truck that is going to be
Define Mine topo file to be updated
connected. The remote port must be defined,
LoadTopo |C:\vmgps\golden csv
as well.
Start Update Stop Update
Generate
Load Contour When the peer terminal is opened, the
Contour
system begins to send local vehicle
Generate Vmnl
Load DTM
DTM
coordinates through the local port, even though
there is not yet a connection to the remote
Figure 62: Peer to Peer comm.
vehicle.
As soon as the Commandl control is executed, the system opens its
communication port to begin receiving position coordinates of the remote vehicle
being monitored. At this point, simultaneous tracking of the local vehicle and the
remote vehicle is achieved.
|
Colorado School of Mines
|
103
With both vehicles now communicating with each other, a remote contour map
update can be performed in real time following the methodology described above in
Section 2, (Real time contouring). The only difference now is that instead of saving the
vehicles own coordinates into a predefined * csv file, remote coordinates coming from
the remote vehicle are stored and then used to generate the survey contour map
representing where the remote vehicle is working at that precise moment.
By defining it in LoadTopo box, the same (x,y,zj file is used to save the local
topographic area (see Section 2, Topo option). Remote data will be then appended to
local survey data, merging together both surveys, local and remote.
This process will eventually allow all mine vehicles to share survey data in real time
with other remote mine vehicles. This gives us the ability not only to monitor vehicle
positions in real time, but also to monitor the evolution of the mine geometry in real
time.
|
Colorado School of Mines
|
105
NIOSH VirtuulMmu
VirtualMine Colorado School of Mines
100
Mapl Map2
Updated Updated 300
Contour DTM
600
Remove
Maps
300
z
Hide Sale Range "X 1200
r Low (• Med < High Viewpoints
Headlight
Start T racking Navigator
Lon j Longitude Eull Screen
Lat CO Latitude
M (2) : Altitude Show Console
x
Plane Dist Preferences
tidckinq
tielp
Topo C:\vmgp8\gold
Ext
Start Update
Generate
Contours VRML Mops TaG rmP iS neJ DaT tr au Bc ak se View Safe Exit
Figure 64: Main interface showing the right click popup menu
to show VRML engine navigational controls
These controls are displayed in the main window as button bars (vertical and
horizontal). The controls are used to change the display view according to user
requirements.
The vertical toolbar contains the controls used to specify navigation type. The
horizontal toolbar contains controls used for predefined actions to change your
position.
Walk, Fly, Study: Three main navigation modes are offered: WALK, FLY, and
STUDY. You can switch the navigation mode by clicking buttons on the vertical
toolbar. Each navigation mode has several options: PLAN, PAN, TURN, and ROLL.
The combination of navigation mode plus option determines camera motion and
|
Colorado School of Mines
|
107
5.8 GUI-Section 5
This section provides the zooming and navigation controls to help the
user track his/her own vehicle as well as to monitor any enabled vehicle
100
nearby.
300
600 Center. Finds and locks the local vehicle at the center of the screen,
creating the effect of leaving the local truck fixed in the center and just
900
moving the background with respect to the truck.
1200
______
Go to. Moves the user point of view directly above the local vehicle at a
distance predefined by the level of zoom.
Center
Zoom Ext. Zooms the loaded screen out far enough to cover the
ISO View extension of all the objects loaded in the system.
Goto
Find. Generates a large bubble of color engulfing the vehicle when the
Zoom user wants to visualize a large section of the mine and at the same time
Ext.
wants to keep tracking the vehicle. Engulfing the vehicle allows it to be
Find pinpointed at almost any zoom level, see Figure 65.
Figure 65: Screen shots showing the Find option in VirtualMine
|
Colorado School of Mines
|
110
CHAPTER VI
6 SYSTEM TESTS AT THE MORENCI MINE, MORENCI, AZ.
6.1 Introduction
VirtualMine needed to be tested in a true open-pit mining operation. Phelps Dodge
provided an invitation to test this system at one of its largest U.S. copper mines,
Morenci.
Phelps Dodge Morenci, Inc., is the largest copper producing operation in North
America. Active operations include an open-pit mine, one concentrator, and two
solution extraction/electro winning (SX/EW) facilities.
Morenci Overview:
e Location: Morenci, Arizona, USA.
• Corporation— 85% interest, Sumitomo Metal Mining Co., Ltd., and 15%
interest Sumitomo Corporation.
• Employees: 2,445
e Total Cu Production—2000: 834 million pounds
• One of the largest Cu mines in the world in terms of production
e Known as one of the best technologically equipped mines
Morenci is recognized as a leader in the application of new technology. Giving
importance and magnitude of Morenci operations, this mine has been chosen to test
various systems including CAT-Trimble and the ModularMining system.
Caterpillar and ModularMining, for example, have implemented and are currently
testing their latest mining systems at Morenci. Thus, it was very important for this
project to carry out tests and to receive feedback from the Morenci experts.
|
Colorado School of Mines
|
113
6.2.1 Instrumentation
Instrumentation was the first step and a very important task to accomplish during
this test. The first challenge was hardware implementation in the mine vehicles, and
setting up the system.
As describe earlier VirtualMine consists of different technologies working together
under the software described in this dissertation:
• Global Positioning System (GPS)
• Radio Wireless Network
• Mobile Computer system and Software.
The software code structure is composed of different technologies as well:
• Central Control Panel based in Visual Basic (VB).
• 3D Graphic Interface based on Virtual Reality Modeling Language (VRML).
• Mapping-contouring Engine based on Surfer™ contouring routine.
6.2.2 Vehicle Tracking and On-Demand Contouring
The second task, after instrumentation, is to test VirtualMine tracking and
contouring capabilities in real-time. In order to run this task, VirtualMine was installed
on a pick-up truck emulating a CAT 797 mine truck.
In order to run this test in the field it was required to import Morenci survey data into
the system. Morenci survey data is based in local state coordinates, thus requiring a
transformation to UTM-GS84.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.