Heuristic for vehicle motion planning with search trees
Peili Guo · KTH Publication Database DiVA (KTH Royal Institute of Technology) · 2020
Motion planning is an essential part of an autonomous system. It finds a sequence of actions for the vehicle to move from start to goal. In this thesis, we study the use of heuristic in the context of motion planning. We propose a heuristic method using a pre-computed heuristic look up table for sampling based motion planning method Rapidly-exploring random trees (RRT). The proposed method is successfully implemented and evaluated in nine different scenarios. We have found improvement in the quality of the path at a short time out in at least six out of the nine scenarios. In certain scenarios, we are able to observe up to 20% reduction of execution time and up to 90% of improvement of the path quality.