Open mkoval opened 8 years ago
I generally agree with this idea. It's especially cool because you could imagine doing things like using a VectorFieldPlanner
to do large extend operations in an CBiRRT with much higher success rates.
Also, this means that SnapPlanner
might take the role of the local planner in most cases, so that we don't have to handle VanDerCorput-sequenced collision checks in every planner and so forth.
@cdellin raised a good point on Reviewable: