Some questions about reboundreplan function

[dssdyx]

I read the code about reboundreplan function and have some confusion: This project doesn't use ESDF map, instead of base point and vector. But it only finds the {p,v} pair within collision segments consisted of control points. Sometimes the generated trajs may be too close to some obstacles, because collision-free optimal problem doesn't care about the segment which is close to obstacles but no happen collision. I want to improve this by increasing inflation. It's effective but not changes this problem fundamentally. I want to know whether your lab try to solve the problem and how to do?

[bigsuperZZZX]

In our perspective, collision avoidance is a strict constraint, and thus should be strictly satisfied by appropriate inflation.