Motivation of Motion Planning
- Robotic motion planning, joints and paths.
- Maintainability studies (separating parts.)
- Digital Actors
Why focus on Narrow Passages
- Computing explicit C-space feasible only up to 3-4 dimensions.
- Randomized planners don't cover narrow passages well.
- Want to improve upon existing techniques:
- Dilating Free Space.
- Sampling Based on Medial Axis
- Sampling near obstacle boundaries.
- ** Need analogy between WS and CS.
Overview of Algorithm
- Efficient - Fast medial axis computation, and QR from distance buffer.
- Simplicity - PRM's are simple and we only modify the sampling technique.
- Narrow passage identification
- Can use gradient to align robots principle axis for biased sampling.
- Based on a heuristic identify some areas to be narrow, and increase samples.
- Global connectivity information - most others don't use.