![Page 1: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/1.jpg)
1
Probabilistic Sensor Models
Introduction to Mobile Robotics
Wolfram Burgard
![Page 2: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/2.jpg)
2
Sensors for Mobile Robots Contact sensors: Bumpers
Proprioceptive sensors Accelerometers (spring-mounted masses) Gyroscopes (spinning mass, laser light) Compasses, inclinometers (earth magnetic field, gravity)
Proximity sensors Sonar (time of flight) Radar (phase and frequency) Laser range-finders (triangulation, tof, phase) Infrared (intensity)
Visual sensors: Cameras
Satellite-based sensors: GPS
![Page 3: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/3.jpg)
3
Proximity Sensors
The central task is to determine P(z|x), i.e., the probability of a measurement z given that the robot is at position x.
Question: Where do the probabilities come from? Approach: Let’s try to explain a measurement.
![Page 4: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/4.jpg)
4
Beam-based Sensor Model Scan z consists of K measurements.
Individual measurements are independent given the robot position.
},...,,{ 21 Kzzzz =
∏=
=K
kk mxzPmxzP
1
),|(),|(
![Page 5: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/5.jpg)
5
Beam-based Sensor Model
∏=
=K
kk mxzPmxzP
1
),|(),|(
![Page 6: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/6.jpg)
6
Typical Measurement Errors of an Range Measurements
1. Beams reflected by obstacles
2. Beams reflected by persons / caused by crosstalk
3. Random measurements
4. Maximum range measurements
![Page 7: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/7.jpg)
7
Proximity Measurement Measurement can be caused by … a known obstacle. cross-talk. an unexpected obstacle (people, furniture, …). missing all obstacles (total reflection, glass, …).
Noise is due to uncertainty … in measuring distance to known obstacle. in position of known obstacles. in position of additional obstacles. whether obstacle is missed.
![Page 8: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/8.jpg)
8
Beam-based Proximity Model Measurement noise
zexp zmax 0
bzz
hit eb
mxzP2
exp )(21
21),|(
−−
=π
η
<
=−
otherwisezz
mxzPz
0e
),|( expunexp
λλη
Unexpected obstacles
zexp zmax 0
![Page 9: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/9.jpg)
9
Beam-based Proximity Model Random measurement Max range
max
1),|(z
mxzPrand η=smallz
mxzP 1),|(max η=
zexp zmax 0 zexp zmax 0
![Page 10: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/10.jpg)
10
Resulting Mixture Density
⋅
=
),|(),|(),|(
),|(
),|(
rand
max
unexp
hit
rand
max
unexp
hit
mxzPmxzPmxzP
mxzP
mxzP
T
αααα
How can we determine the model parameters?
![Page 11: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/11.jpg)
11
Raw Sensor Data Measured distances for expected distance of 300 cm.
Sonar Laser
![Page 12: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/12.jpg)
12
Approximation Maximize log likelihood of the data
Search space of n-1 parameters. Hill climbing Gradient descent Genetic algorithms …
Deterministically compute the n-th parameter to satisfy normalization constraint.
)|( expzzP
![Page 13: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/13.jpg)
Approximation Results
Sonar
Laser
300cm 400cm
![Page 14: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/14.jpg)
14
Example
z P(z|x,m)
![Page 15: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/15.jpg)
16
Discrete Model of Proximity Sensors
Instead of densities, consider discrete steps along the sensor beam.
Laser sensor Sonar sensor
![Page 16: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/16.jpg)
17
Approximation Results
Laser
Sonar
![Page 17: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/17.jpg)
18
"sonar-0"
0 10 20 30 40 50 60 700 102030405060700
0.050.1
0.150.2
0.25
Influence of Angle to Obstacle
![Page 18: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/18.jpg)
19
"sonar-1"
0 10 20 30 40 50 60 700 102030405060700
0.050.1
0.150.2
0.250.3
Influence of Angle to Obstacle
![Page 19: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/19.jpg)
20
"sonar-2"
0 10 20 30 40 50 60 700 102030405060700
0.050.1
0.150.2
0.250.3
Influence of Angle to Obstacle
![Page 20: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/20.jpg)
21
"sonar-3"
0 10 20 30 40 50 60 700 102030405060700
0.050.1
0.150.2
0.25
Influence of Angle to Obstacle
![Page 21: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/21.jpg)
22
Summary Beam-based Model Assumes independence between beams.
Justification? Overconfident!
Models physical causes for measurements. Mixture of densities for these causes. Assumes independence between causes. Problem?
Implementation Learn parameters based on real data. Different models should be learned for different angles at
which the sensor beam hits the obstacle. Determine expected distances by ray-tracing. Expected distances can be pre-processed.
![Page 22: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/22.jpg)
23
Scan-based Model Beam-based model is … not smooth for small obstacles and at edges. not very efficient.
Idea: Instead of following along the beam, just check the end point.
![Page 23: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/23.jpg)
24
Scan-based Model Probability is a mixture of … a Gaussian distribution with mean at distance to
closest obstacle, a uniform distribution for random
measurements, and a small uniform distribution for max range
measurements. Again, independence between different
components is assumed.
![Page 24: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/24.jpg)
25
Example
P(z|x,m)
Map m
Likelihood field
![Page 25: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/25.jpg)
26
San Jose Tech Museum
Occupancy grid map Likelihood field
![Page 26: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/26.jpg)
27
Scan Matching Extract likelihood field from scan and use it
to match different scan.
![Page 27: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/27.jpg)
29
Properties of Scan-based Model
Highly efficient, uses 2D tables only. Distance grid is smooth w.r.t. to small
changes in robot position.
Allows gradient descent, scan matching.
Ignores physical properties of beams.
![Page 28: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/28.jpg)
30
Additional Models of Proximity Sensors
Map matching (sonar, laser): generate small, local maps from sensor data and match local maps against global model.
Scan matching (laser): map is represented by scan endpoints, match scan into this map.
Features (sonar, laser, vision): Extract features such as doors, hallways from sensor data.
![Page 29: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/29.jpg)
31
Landmarks
Active beacons (e.g., radio, GPS) Passive (e.g., visual, retro-reflective) Standard approach is triangulation
Sensor provides distance, or bearing, or distance and bearing.
![Page 30: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/30.jpg)
32
Distance and Bearing
![Page 31: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/31.jpg)
33
Probabilistic Model 1. Algorithm landmark_detection_model(z,x,m):
2.
3.
4.
5. Return
22 ))(())((ˆ yimximd yx −+−=
),ˆprob(),ˆprob(det αεααε −⋅−= dddp
θα ,,,,, yxxdiz ==
θα −−−= ))(,)(atan2(ˆ ximyim xy
detp
![Page 32: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/32.jpg)
34
Distributions
![Page 33: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/33.jpg)
35
Distances Only No Uncertainty
P1 P2 d1 d2
x
X’
a
)(
2/)(22
1
22
21
2
xdy
addax
−±=
−+=
P1=(0,0) P2=(a,0)
![Page 34: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/34.jpg)
36
P1
P2
D1
z1
z2 α
P3 D2 β
z3
D3
Bearings Only No Uncertainty
P1
P2 D1
z1
z2
α
α
αcos2 2122
21
21 zzzzD −+=
)cos(2)cos(2)cos(2
2123
21
23
2123
22
22
2122
21
21
βα
β
α
+−+=
−+=
−+=
zzzzDzzzzDzzzzDLaw of cosine
![Page 35: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/35.jpg)
37
Bearings Only With Uncertainty
P1
P2
P3
P1
P2
Most approaches attempt to find estimation mean.
![Page 36: Introduction to Mobile Robotics Probabilistic Sensor Modelsais.informatik.uni-freiburg.de/teaching/ss17/robotics/slides/07... · Standard approach is triangulation Sensor provides](https://reader031.vdocument.in/reader031/viewer/2022022519/5b14f0ca7f8b9a201a8cc21c/html5/thumbnails/36.jpg)
38
Summary of Sensor Models Explicitly modeling uncertainty in sensing is key to
robustness. In many cases, good models can be found by the
following approach: 1. Determine parametric model of noise free measurement. 2. Analyze sources of noise. 3. Add adequate noise to parameters (eventually mix in
densities for noise). 4. Learn (and verify) parameters by fitting model to data. 5. Likelihood of measurement is given by “probabilistically
comparing” the actual with the expected measurement. This holds for motion models as well. It is extremely important to be aware of the
underlying assumptions!