Probabilistic Roadmap

1
Probabilistic Roadmap

• Matt Troup
• Elie Shammas

2
Introduction

• Preprocessing
• Construct a probabilistic roadmap by adding
  randomly distributed points in the free space and
  connecting them via a fast local planner
• Query
• Connect the start and goal configuration to the
  roadmap and find a path in the roadmap

3
Components

• Roadmap
• Undirected graph
• Nodes configurations
• Edges paths
• qinit, qinit
• qgoal, qgoal

4
Planners

• Random Bounce
• Local Planner
• Strong Planner

5
Phases Preprocessing

• Roadmap construction

6
Phases Preprocessing

• Roadmap construction

7
Phases Preprocessing

• Roadmap construction

8
Phases Preprocessing

• Roadmap construction

9
Phases Preprocessing

• Roadmap construction

10
Phases Preprocessing

• Roadmap expansion

11
Phases Preprocessing

• Roadmap expansion

12
Phases Preprocessing

• Component Reduction

13
Phases Preprocessing

• Final Roadmap

14
Phases Query

• Final Roadmap

qinit
qgoal
qgoal
qinit
15
Phases Query

• Final Roadmap

16
Improvements

• Interweaving preprocessing and query phases.
• Time limit for strong planner
• Improving local planner for non-point robots
• Path Trimming

17
Phases Query

• Final Roadmap

18
Results
19
Conclusion

• Fast
• Probabilistic completeness
• For 316 dofs in static environments