The Ruckig trajectory smoother currently stretches the entire trajectory in time until all kinematic limits are obeyed. This works but it can slow the trajectory more than necessary. It's better to feed in just a few waypoints at once so only the waypoints needing it get stretched.

The downside of passing just a few waypoints at once is very long runtimes if the trajectory is very large. So, I think the number of waypoints passed to Ruckig should be chosen in a smarter way.

Attached is some code to feed in a configurable number of waypoints.

process_waypoints_in_chunks.txt

A good way to select the number of waypoints is probably:

• The smaller of (100 waypoints, 10% of waypoints)