Jiangfan Li
Monday, July 1, 2019 | 1 minute
Exploration and Implementation of Motion Planning Algorithm with Specified Optimization Goal
Jul 2019 - Sep 2019
Aqrose Technology, Beijing, China
- Reviewed and synthesized literature on variances of RRT* algorithms, posted a tutorial session on video blog.
- Implemented a modified BIT* Algorithm using OMPL in C++ on the trajectory planner module of a Fanuc robot to avoid collision; tested on OpenRave; achieved an increase of the average clearance by 100% with less planning time.
Some Sample-based Planning Algorithms
- Source Code
- RRT
- biRRT
- RRT*
- BIT*
- biBIT*
BIT* with clearance penalty
A introduction to BIT* algorithm and the modified version with clearance penalty
A case with ABB IRB6700 in a stacking station:
| Planner Name | Planning Time(s) | Trajectory waypoint number | Cost (in joint angle distance) | min. Clearance (m) | avg. Clearance (m) |
|---|---|---|---|---|---|
| BITstarDis(My Implementation) | 2.34 | 3 | 3.04 | 0.14 | 0.26 |
| BITstar | 2.51 | 5 | 2.72 | 0.05 | 0.20 |
| InformedRRTstar | 2.45 | 5 | 4.66 | 0.06 | 0.25 |
| RRTConnect | 0.80 | 7 | 10.41 | 0.10 | 0.29 |
Trajectories:
BITstarDis
BITstar
InformedRRTstar
RRTConnect
With the new variant of BIT* planner we proposed, BIT*Dis, the trajectory is more smooth with less trajectory points and a higher minimal clearance from the obstacles.
Another example trajectory generated by BIT*Dis:
comments powered by Disqus