Required Reading…

Constrained Planning Using Manifolds

From the Kavraki Lab

From the MoveIt Tutorials


Two-Stage Planning: OMPL + Trajectory Optimization

The basic idea here is to first generate plan with OMPL, then to use a trajectory-based planner like CHOMP, STOMP or TrajOpt to improve it, and apparently it’s pretty fast.

Pablo’s feeling, is that if we write something like this, we should go with TrajOpt, and use this code as a template…

Planning for Multiple Goals

I think we want to..

  • approach a table…
  • scan from left to right, in rows from front to back, and pick up all the white blocks that we can
  • if we try and fail, then we move on, till we’ve covered all we can from our current nav position
  • then move to another Nav position (say the other side of the table)
  • and repeat.