09d ekf localization
TRANSCRIPT
-
8/10/2019 09d Ekf Localization
1/38
Wolfram Burgard, Cyrill Stachniss,
Maren Bennewitz, Kai Arras
EKF Localization
Introduction toMobile Robotics
Slides by Kai Arras and Wolfram Burgard
Last update: June 2010
-
8/10/2019 09d Ekf Localization
2/38
Localization
Given
Map of the environment.
Sequence of sensor measurements.
Wanted Estimate of the robots position.
Problem classes
Position tracking
Global localization
Kidnapped robot problem (recovery)
Using sensory information to locate the robotin its environment is the most fundamentalproblem to providing a mobile robot withautonomous capabilities. [Cox 91]
-
8/10/2019 09d Ekf Localization
3/38
Landmark-based Localization
EKF Localization:Basic Cycle
3
-
8/10/2019 09d Ekf Localization
4/38
Landmark-based Localization
EKF Localization:Basic Cycle
4
-
8/10/2019 09d Ekf Localization
5/38
Landmark-based Localization
EKF Localization:Basic Cycle
5
raw sensory data
landmarks
innovation from
matched landmarks
predictedmeasurements in
sensor coordinates
landmarks in globalcoordinates
encoder measurements
predicted
state
posteriorstate
-
8/10/2019 09d Ekf Localization
6/38
State Prediction (Odometry)
Landmark-based Localization
6
Control uk: wheel displacementssl, sr
Error model: linear growth
Nonlinear process model f:
-
8/10/2019 09d Ekf Localization
7/38
State Prediction (Odometry)
Landmark-based Localization
7
Control uk: wheel displacementssl, sr
Error model: linear growth
Nonlinear process model f:
-
8/10/2019 09d Ekf Localization
8/38
Landmark-based Localization
Landmark Extraction (Observation)
8
Extractedlines
Hessian line model
Extracted linesin model space
Raw laserrange data
-
8/10/2019 09d Ekf Localization
9/38
Landmark-based Localization
Measurement Prediction
...is a coordinate frame transform world-to-sensor
Given the predicted state (robot pose),predicts the location and location
uncertainty of expected
observations in sensor coordinates
9
model space
-
8/10/2019 09d Ekf Localization
10/38
Data Association (Matching) Associates predicted measurements
with observations
Innovationand innovation
covariance
Matching onsignificance
level alpha
Landmark-based Localization
10
model space
Green: observation
Magenta: measurement prediction
-
8/10/2019 09d Ekf Localization
11/38
Landmark-based Localization
Update
Kalman gain
State update (robot pose)
State covariance update
11
Red: posterior estimate
-
8/10/2019 09d Ekf Localization
12/38
EKF Localization with Point Features
Landmark-based Localization
-
8/10/2019 09d Ekf Localization
13/38
1. EKF_localization ( t-1,!t-1, ut, zt,m):
Prediction:
2.
3.
4.
5.
6. "t =Gt"t#1GtT+ B
tQ
tB
t
T
Bt ="g(ut,t#1)
"ut=
"x'
"v t
"x'
"$t"y'
"v t
"y'
"$t"%'
"v t
"%'
"$t
&
'
((((((
)
*
++++++
Qt=
"1| v
t |+"
2 |#
t |( )
2
0
0 "3 | vt |+"4 |#t |( )
2
$
%&
&
'
()
)Motion noise
Jacobian of gw.r.t location
Predicted mean
Predicted covariance
Jacobian of gw.r.t control
-
8/10/2019 09d Ekf Localization
14/38
1. EKF_localization ( t-1,!t-1, ut, zt,m):
Correction:
2.
3.
4.
5.
6.
7.
8.
St= H
t"
tH
t
T+ R
t
Rt=
"r
20
0 "r
2
#
$%
&
'(
Predicted measurement mean
Innovation covariance
Kalman gain
Updated mean
Updated covariance
Jacobian of hw.r.t location
-
8/10/2019 09d Ekf Localization
15/38
EKF Prediction Step
-
8/10/2019 09d Ekf Localization
16/38
EKF Observation Prediction Step
-
8/10/2019 09d Ekf Localization
17/38
EKF Correction Step
-
8/10/2019 09d Ekf Localization
18/38
Estimation Sequence (1)
-
8/10/2019 09d Ekf Localization
19/38
Estimation Sequence (2)
-
8/10/2019 09d Ekf Localization
20/38
Comparison to GroundTruth
-
8/10/2019 09d Ekf Localization
21/38
[Arras et al. 98]:
Laser range-finder and vision
High precision (
-
8/10/2019 09d Ekf Localization
22/38
EKF Localization Example
Line and point landmarks
-
8/10/2019 09d Ekf Localization
23/38
EKF Localization Example
Line and point landmarks
-
8/10/2019 09d Ekf Localization
24/38
EKF Localization Example
Expo.02: Swiss National Exhibition 2002
Pavilion "Robotics"
11 fully autonomous robots tour guides, entertainer, photographer 12 hours per day 7 days per week
5 months
3,316km travel distance almost 700,000 visitors 400 visitors per hour
Localization method: Line-Based EKF
-
8/10/2019 09d Ekf Localization
25/38
EKF Localization Example
-
8/10/2019 09d Ekf Localization
26/38
26
Global EKF Localization
Interpretation tree
-
8/10/2019 09d Ekf Localization
27/38
27
Global EKF Localization
Env. Dynamics
-
8/10/2019 09d Ekf Localization
28/38
28
Global EKF Localization
Geometric constraints we can exploit
Location independent constraints
Unary constraint:intrinsic property of feature
e.g. type, color, size
Binary constraint:
relative measure between featurese.g. relative position, angle
All decisions on a significance level "
-
8/10/2019 09d Ekf Localization
29/38
29
Global EKF Localization
Interpretation Tree[Grimson 1987], [Drumheller 1987],
[Castellanos 1996], [Lim 2000]
Algorithm
backtracking
depth-first
recursive
uses geometric constraints
worst-case exponentialcomplexity
-
8/10/2019 09d Ekf Localization
30/38
30
Global EKF Localization
Pygmalion
"= 0.95 , p= 2
-
8/10/2019 09d Ekf Localization
31/38
31
Global EKF Localization
"= 0.95 , p= 3
Pygmalion
-
8/10/2019 09d Ekf Localization
32/38
32
Global EKF Localization
"= 0.95 , p= 4texe:633 msPowerPC at 300
MHz
Pygmalion
-
8/10/2019 09d Ekf Localization
33/38
33
Global EKF Localization
"= 0.95 , p= 5
texe:633 ms(PowerPC at 300 MHz)
Pygmalion
-
8/10/2019 09d Ekf Localization
34/38
05.07.02, 17.23 h
Global EKF Localization
"= 0.999
At Expo.02
[Arras et al. 03]
-
8/10/2019 09d Ekf Localization
35/38
texe= 105 ms
05.07.02, 17.23 h
Global EKF Localization
"= 0.999
At Expo.02
[Arras et al. 03]
-
8/10/2019 09d Ekf Localization
36/38
05.07.02, 17.32 h
Global EKF Localization
"= 0.999
At Expo.02
[Arras et al. 03]
-
8/10/2019 09d Ekf Localization
37/38
05.07.02, 17.32 h
Global EKF Localization
"= 0.999 texe= 446 ms
At Expo.02
[Arras et al. 03]
-
8/10/2019 09d Ekf Localization
38/38
EKF Localization Summary
EKF localization implements pose tracking Very efficient and accurate
(positioning error down to subcentimeter)
Filter divergence can cause lost situations from
which the EKF cannot recover Industrial applications
Global EKF localization can be achieved using
interpretation tree-based data association Worst-case complexity is exponential
Fast in practice for small maps