CSE-473

1
CSE-473
Mobile Robot Mapping
2
Mapping with Raw Odometry
3
Why is SLAM a hard problem?
SLAM robot path and map are both unknown
Robot path error correlates errors in the map
4
Why is SLAM a hard problem?
Robot pose uncertainty

• In the real world, the mapping between
  observations and landmarks is unknown
• Picking wrong data associations can have
  catastrophic consequences
• Pose error correlates data associations

5
Graphical Model of Mapping
u
u
u
1
0
t-1
...
x
x
x
x
2
t
1
0
m
z
z
z
2
t
1
6
(E)KF-SLAM

• Map with N landmarks(32N)-dimensional Gaussian
• Can handle hundreds of dimensions

7
EKF-SLAM
Map Correlation matrix
8
EKF-SLAM
Map Correlation matrix
9
EKF-SLAM
Map Correlation matrix
10
Victoria Park Data Set
courtesy of E. Nebot
11
Victoria Park Data Set Vehicle
courtesy of E. Nebot
12
Data Acquisition
courtesy of E. Nebot
13
Estimated Trajectory
courtesy of E. Nebot
14
Graph-SLAM Idea
15
Mapping the Allen Center
16
Comparison to Ground Truth Map
17
Three Mapping Runs
18
Three Overlayed Maps
19
3D Outdoor Mapping
108 features, 105 poses, only few secs using cg.
20
Map Before Optimization
21
Map After Optimization
22
Autonomous Navigation
Courtesy of W. Burgard
23
Rao-Blackwellized Mapping
Compute a posterior over the map and possible
trajectories of the robot
map and trajectory
measurements
robot motion
map
trajectory
24
FastSLAM
Robot Pose
2 x 2 Kalman Filters

Particle M
Begin courtesy of Mike Montemerlo
25
Example
3 particles
26
Rao-Blackwellized Mapping with Scan-Matching
Map Intel Research Lab Seattle
27
Frontier Based Exploration
Yamauchi et al. 96, Thrun 98
28
(No Transcript)