View
217
Download
0
Category
Tags:
Preview:
Citation preview
On Improving the Clearance for Robots in High-Dimensional Configuration Spaces
Roland Geraerts and Mark Overmars
IROS 2005
Outline
• Related work
• Retraction algorithm
• Algorithmic details
• Experiments
• Conclusions
Related Work
• Preprocessing techniques– Extract path from graph with cycles– Create path along the medial axis
• Post processing technique– Retract path to the medial axis (MA)
d
d
Related Work
• Preprocessing techniques– Extract path from graph with cycles– Create path along the medial axis
• Post processing technique– Retract path to the medial axis (MA)
Related Work
• Preprocessing techniques– Extract path from graph with cycles– Create path along the medial axis
• Post processing technique– Retract path to the medial axis (MA)
Initial path Retraction to MA Retraction to ridges
Retraction Algorithm
Algorithmic Details – Metric
Algorithmic Details – Vector
Algorithmic Details – Validate
maximumstep size
Experimental Setup
• Pentium 4 (2.66GHz)
• Comparison of retraction techniques– Workspace (1 run)– Configuration space (100 runs)
• Environments / paths
planar (2 dofs) hole (6 dofs) manipulator (6 dofs)
Experimental Results
min avg time
initialworkspaceC-space
0.351.801.79
2.844.494.47
0.01.5
15.0
initial paths workspace C-space
n.a.
Clearance
min avg time
initialworkspaceC-space
0.210.441.27
1.553.374.69
0.01.5
50.0
min avg time
initialworkspaceC-space
0.01 n.a.0.16
0.13 n.a.0.33
0.0 n.a.60.0
Conclusions
• Clearance along paths is improved– No complicated data structures are needed– The algorithm can handle robots with an
arbitrary amount of DOFs
• Running times are too high for online usage– Does not matter for high-cost environments– Smarter implementation decreased the running
times by an order of magnitude
Current Work
• Retraction of a roadmap yields online performance
Reachability roadmap Retracted roadmap
Current Work
• Retraction of a roadmap yields online performance
Reachability roadmap Retracted roadmap
Recommended