I've developed an open-source motion planner that achieves sub-millisecond planning times for moderately complex problems with articulated robots. Initially developed as a general robotics programming framework including IK and trajectory optimization, in addition to motion planning, I ended up focusing on optimizing the motion planning component just for fun.
The performance improvements came through persistent tuning using perf profiling. While it doesn't match VAMP's performance (the world's undisputedly fastest motion planner as of 2024), I think it's interesting that persistent tuning without any groundbreaking innovations still achieved sub-millisecond planning times.
Show HN: Sub-Millisecond Motion Planner for Articulated Robots | Heykuki News