MTP FY03 Year End Review – Oct 20-24, 2003 - 1 Visual Odometry Yang Cheng Machine Vision Group Section 348 ycheng@jpl.nasa.govycheng@jpl.nasa.gov Phone:
Post on 13-Jan-2016
212 Views
Preview:
Transcript
MTP FY03 Year End Review – Oct 20-24, 2003 - 1
Visual Odometry
Yang ChengMachine Vision Group
Section 348ycheng@jpl.nasa.gov Phone: 4-1857
12/16/2003
MTP FY03 Year End Review – Oct 20-24, 2003 - 2
Outline of this Talk
• Brief History• Algorithm• Software structure and interface• Software Features• Ground truth measurement• Some results• Future works
MTP FY03 Year End Review – Oct 20-24, 2003 - 3
Brief History
• H. Moravec’s PhD Thesis, “ Obstacle Avoidance and Navigation in the Real World by a Seeing Robot Rover, Stanford University, 1980
• Larry Matthies’ PhD Thesis, “Dynamic Stereo Vision”, Oct, 1989, CMU.
• A version of Visual Odometry in C was implemented in early 1990s in JPL.
• A C++ version of visual odometry was implemented by MTP Slope Navigation task led by Larry Matthies in 2001.
• The visual odometry has been ported to CLARAty and demonstrated onboard motion estimation on Rock 8 in 2002.
• The visual odometry has been used successfully on slip compensation by the slope navigation task.
• The visual odometry has been integrated officially to MER navigation software and demonstrated successfully in 2003.
• A few other versions of visual odometry were developed in academic and industry communities.
MTP FY03 Year End Review – Oct 20-24, 2003 - 4
Visual Odometry
Feature Selections
Feature Stereo Matching
Feature Gap Analysis
Feature Tracking
Rigidity Test
Least Medians SquareSchonemann Motion Estimation
Maximum Likelihood Motion Estimation
Visual Odometry Fusion
Input Images Input Motion
Output Motion
To use a (stereo) image sequence to track 3-D point features, or landmark, to estimate the motion of the vehicle.
MTP FY03 Year End Review – Oct 20-24, 2003 - 5
Feature (Landmark) Selection
• A landmark is a patch of image which must exhibit intensity variation that allows the landmark to be localized in subsequent image.
Input Image
LandmarksInterest Image
Forstner operator
MTP FY03 Year End Review – Oct 20-24, 2003 - 6
Feature Stereo Matching (Pyramid Searching)
Left Image
Right Image
MTP FY03 Year End Review – Oct 20-24, 2003 - 7
Feature Gap Analysis and Triangulation Error
Gap
Gap Analysis
Cameras
Error Vs Location Error Vs Correlation Peak
Error Vs Ray Gap
L C.
R.C.
L.C.
R.C.
Gaussian Error Model
MTP FY03 Year End Review – Oct 20-24, 2003 - 8
Feature Tracking
Left Image
Right Image
Feature pred
iction
Qp
Qc
Previous Frame
Current Frame
MTP FY03 Year End Review – Oct 20-24, 2003 - 9
Motion Estimation (Least-Squares Vs Maximum Likelihood)
• A closed form solution
• Rotation, R, with orthogonal constrain is estimated first
• Translation, T, is then estimated.
• Reflect the quality of the observations.
• It is fast.
• The resulting motion estimates can be substantially inferior.
• An nonlinear optimization solution
• Fully reflects the error model
• It is relative slow
• It needs an initial estimate.
• It is sensitive to outliers
• Its motion estimates in general is much superior than the least-squares estimation.
Least-squares Estimation Maximum Likelihood Estimation
ipici vTRQQ
ii
Tii
pici
eewTRq
TRQQe
),(
MTP FY03 Year End Review – Oct 20-24, 2003 - 10
Least-squares Estimation
ii
Tii
pici
eewTRq
TRQQe
),(
jTi
i jiii
Tiij
Tjiii
jTii
Ti
rrmrrleewmlTRq
jijirrrr
3
1
3
0,
)1(}{),,,(
}3,2,1{,01
Merit Function:
Orthogonal constrains:
Solutions:
][1
1
21
21
21
RQQw
TUVR
USVEQQw
AEQQwA
QQQQww
T
TTTcipii
picii
MTP FY03 Year End Review – Oct 20-24, 2003 - 11
Maximum Likelihood Estimation
Merit Function: iTiWeeM
W = covariance matrix of the feature i
Solutions: To linearize the merit function and determine the three attitude and three translation iteratively. Page 150 of Larry Matthies’ thesis
MTP FY03 Year End Review – Oct 20-24, 2003 - 12
Visual Odometry Interface
VOMotionStart( leftCam, rightCam, ParameterFile, leftImage, rightImage, leftDisp, InitialMotion)
VOMotion(leftImage, rightImage, leftDisp, InitialMotion, *estMotion)
Camera models: CAHV, CAHVOR, CAHVORE
Motion file: Position[3], attitude [3], covarence[6][6]
leftDisp: the disparity image generated by stereo processing.
Parameter File contains 48 parameters
MTP FY03 Year End Review – Oct 20-24, 2003 - 13
Some VO ParametersVO_MAX_NUM_VO_FEATURES 600 features
VO_MIN_NUM_VO_FEATURES 8 iteration
VO_VO_MAX_PIXEL_OFFSET 1 pixel
VO_MAX_VO_ITERATIONS 50 iteration
VO_VO_CORR_WINDOW_ROWS 9 pixel
VO_VO_CORR_WINDOW_COLS 9 pixel
VO_VO_TRACK_WINDOW_SIZE 50 pixel
VO_VO_SELECT_WINDOW_SIZE 9 pixel
VO_VO_NUM_IMAGE_PAIRS 4 images
VO_VO_IMAGE_ROWS 640 pixel
VO_VO_IMAGE_COLS 480 pixel
VO_SCHONEMANN_ITERATIONS 50 iteration
VO_VO_MIN_DIST_FEATURE 0.5 meter
VO_VO_MAX_DIST_FEATURE 20.0 meter
VO_VO_AFFINE_MATCH_FLAG 0 Boolean
VO_MAX_DELTA 0.000006
VO_DEFAULT_VO_MIN_CORRELATION 0.8 correlation
MTP FY03 Year End Review – Oct 20-24, 2003 - 14
Arroyo Data Collection
1. About 8 meters of image (20 cm step) sequence were collected at JPL arroyo in March, 2002.2. Onboard IMU, wheel odometry and other data were collected.3. Ground truth data (position and attitude) were collectedby totalstation.
MTP FY03 Year End Review – Oct 20-24, 2003 - 15
Semiautomatic Rover Position and Attitude Measurement
Pitch, Roll, Heading error < 0.5 degree;Position error < 3mm.
1
2
3
Three points are measured at each stop.
The position and attitude can be determined.
Total Station & Prism
MTP FY03 Year End Review – Oct 20-24, 2003 - 16
Motion Estimation (X)
0
1
2
3
4
5
6
7
8
9
0 5 10 15 20 25 30 35 40 45 50
Image Step
X (
m)
VO (Rear)
WO
VO ( front)
GT
MTP FY03 Year End Review – Oct 20-24, 2003 - 17
Motion Estimation (Y)
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
0 5 10 15 20 25 30 35 40 45 50
Image Step
Y (
m)
VO (rear)
VO (front)
GT
WO
MTP FY03 Year End Review – Oct 20-24, 2003 - 18
Motion Estimation (Z)
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0 5 10 15 20 25 30 35 40 45 50
Image Step
Z(m
)
VO(rear)
WO
VO( front)
GT
MTP FY03 Year End Review – Oct 20-24, 2003 - 19
Heading Estimation
-15
-10
-5
0
5
10
15
20
0 5 10 15 20 25 30 35 40 45 50
Image Step
Hea
din
g (
deg
)
VO(Rear)
VO (Front)
GT
MTP FY03 Year End Review – Oct 20-24, 2003 - 20
Roll Estimation
-5
0
5
10
15
20
25
0 5 10 15 20 25 30 35 40 45 50
Image Step
Rx(
Deg
) Rx(Rear)
Rx(Front)
Rx(GT)
MTP FY03 Year End Review – Oct 20-24, 2003 - 21
-5
0
5
10
15
20
0 5 10 15 20 25 30 35 40 45 50
Image Step
RY
(deg
) VO(Rear)
VO(Front)
GT
Pitch Estimation
MTP FY03 Year End Review – Oct 20-24, 2003 - 22
VO Fusion (front and rear Has Camera )
-0.02
-0.015
-0.01
-0.005
0
0.005
0.01
0.015
0.02
0.025
0.03
0 5 10 15 20 25 30 35 40 45 50
Rear Rear + front
Absolute Error (x)
MTP FY03 Year End Review – Oct 20-24, 2003 - 23
VO Fusion (front and rear Has Camera )
Image Step
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0 5 10 15 20 25 30 35 40 45 50
Ab
solu
te E
rro
r (m
)
Rear Rear + front
Absolute Error (Y)
MTP FY03 Year End Review – Oct 20-24, 2003 - 24
VO Fusion (front and rear Has Camera )
Comparison Between Single VO and Fusioned VO
-0.14
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0 5 10 15 20 25 30 35 40 45 50
Image Step
Ab
so
lute
Err
or
(m)
Rear Rear + front
Absolute Error (Z)
MTP FY03 Year End Review – Oct 20-24, 2003 - 25
Absolute Error (Heading)
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 5 10 15 20 25 30 35 40 45 50
Image Step
He
ad
ing
Err
or
Rear Rear + front
MTP FY03 Year End Review – Oct 20-24, 2003 - 26
Absolute Error (Pitch)
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 5 10 15 20 25 30 35 40 45 50
Image Step
Pit
ch
Ab
olu
te E
rro
r (d
eg
)
Rear Rear + front
MTP FY03 Year End Review – Oct 20-24, 2003 - 27
Absolute Error (Roll)
Roll Error
-1.5
-1
-0.5
0
0.5
1
1.5
0 5 10 15 20 25 30 35 40 45 50
Image Step
Ab
olu
te E
rro
r (d
eg
)
Rear Rear + front
MTP FY03 Year End Review – Oct 20-24, 2003 - 28
Heading Estimation
0 4
7 10
MTP FY03 Year End Review – Oct 20-24, 2003 - 29
0
5
10
15
20
25
30
35
1 2 3 4 5 6 7 8 9
Steps
He
ad
ing
(de
g)
VO(Rear)
VO(Front)
GT
Heading Estimation (1)
MTP FY03 Year End Review – Oct 20-24, 2003 - 30
05
1015
20253035
1 2 3 4 5 6 7 8 9
Image Step
Hea
din
g (
deg
)
Small Angle Steering Large Angle Steering
Heading Estimation (2)
MTP FY03 Year End Review – Oct 20-24, 2003 - 31
Field Test
Location• Johnson Valley, Mojave Desert, CA• Sandy slopes of up to 20-25° slopes
Logistics• 4 days – 4 people
– 1.5 days of setup and break down
– 2.5 days of experimentation
Motivation• Mars Yard is too small and
has no slopes– The size is mostly a factor for
visual odometry which looks far beyond traverse distance
MTP FY03 Year End Review – Oct 20-24, 2003 - 32
Sample of images
MTP FY03 Year End Review – Oct 20-24, 2003 - 33
Field Test Results
Visual Odometry vs. Ground Truth
0 5 10 15 20 25
-4
-2
0
2
x (meters)
y (m
ete
rs)
ground truthvisual odometry
area expanded (and rotated) in next slide
• Error (0.37 m) is less than 1.5% of distance traveled (29 m)
• Ground truth data collected with a Leica Total Station and four rover mounted prisms
MTP FY03 Year End Review – Oct 20-24, 2003 - 34
Field Test Results
Slip Compensation/Path Following Results
carrot heading
visual odometry pose
kinematics pose
desired path
0 2 4 6 8 10
0
0.5
1
1.5
2
x (meters)
y (m
eter
s)
expanded below
5 5.5 6 6.5 7 7.5 81.2
1.4
1.6
1.8
2
2.2
x (meters)
y (m
eter
s)
• There is a noticeable bias between the visual odometry pose and the kinematics pose in the y direction of many estimates; this is due to the downhill slippage of the rover; this bias is being compensated for in the slip compensation algorithm
MTP FY03 Year End Review – Oct 20-24, 2003 - 35
MER VO Test (Rough)
MTP FY03 Year End Review – Oct 20-24, 2003 - 36
MER Test
MTP FY03 Year End Review – Oct 20-24, 2003 - 37
Target Approach
(a)
Target
(b)
Designated Target
Target Tracking
time = t2(avoiding an obstacle)
time = t1
1st Frame 37th Frame after 10 m
MTP FY03 Year End Review – Oct 20-24, 2003 - 38
Integrated 2D/3D Tracker
Not tested onboard in the integrated system
Stereo ProcessStereo Process
IMU
Hazard Cameras
NavigatorNavigator
R8 LocomotorR8 Locomotor
Motion Cmd
Navigator
DepthMap
Harris Multiple Feature Extractor
Harris Multiple Feature Extractor
Rover Pose Estimate
Visual Odometry
Visual OdometryNon-flat Surface
Filter
Non-flat Surface Filter
Rover Pose Estimate + uncertainty?
Wheel odo
Compute MastPointing Angle
Compute MastPointing Angle
Tilt
Sub windowStereo Map
Sub windowStereo Map
Pan & Tilt Angles
Large Uncert-ainty?
Large Uncert-ainty?
DesignateTarget (DT)(r,c) in right image
No
DT(r,c) to DT (x,y,z)
Mast Cameras
Yes
Output: DT(r,c) at t0+ Δt
Mast ImagesDepth Disparity
Expected2D location
Pan-TiltController
Pan-TiltController
Mast KinematicsMast KinematicsDesignated Point Visual Tracking• Track once w/ 4 mm camera• Seed & track again w/ 16 mm camera
Pointing Vector
1st Tier 2D location estimate
1st Tier 2D location estimate
Adaptive View-Based Matching
Adaptive View-Based Matching
Normalized Cross Correlation
Normalized Cross Correlation
Surface Normals Grow Feature Win
Surface Normals Grow Feature Win
Single Pt Stereo
Single Pt Stereo
DesignatedTarget
Template
2nd Tier 2D location estimate
Affine TrackerAffine Tracker
MTP FY03 Year End Review – Oct 20-24, 2003 - 39
Tracking Results over Rough Terrain
Tracking Video
View from 4 mm camera
View from 16 mm camera
MTP FY03 Year End Review – Oct 20-24, 2003 - 40
Ground Truth Data Collection System
• Automatically tracks the position of 1 prism and finds the 3 other prisms when rover stops
• Simplifies and speeds the collection of ground truth data in field tests
• Locates rover frame in world frame and the initial rover frame
• +/- 2mm position accuracy
• +/- 0.3º orientation accuracy
MTP FY03 Year End Review – Oct 20-24, 2003 - 41
Future Works
• A real-time Visual Odometry• Data Fusion with other sensors (IMU …) to achieve better
estimation• Visual Odometry Applications
top related