Research On Path Planning Algorithms For Au-tonomous Land Robot

Autonomous land robot has become a new research hotspot of intelligent robot field. It covers the frontier of intelligent control, pattern recognition, machine learning. Path planning is one of the key technologies of the autonomous land robot. Almost every task’s implementation is related to path planning.The research of path planning has important theoretical and practical significance.Continuous A*algorithm is improved in this thesis. The improved algorithm can adaptively set the control granularity. In order to improve the search efficiency, weighted Continuous A*algorithm is proposed in this thesis. To avoid local minima, a randomized version of the Continuous A*algorithm is proposed in this thesis, called RC*algorithm. By randomly generating a lot of local objectives, the RC*algorithm can effectively solve the local minima problems. In addition, when the local path planning is finished, the RC*algorithm can release memory resources timely, so that RC*algorithm can be extended to the field of large-scale path planning.R*GPU algorithm is implemented in this thesis. R*GPU algorithm is a typical rep-resentative of GPU accelerated path planning. By sampling states randomly,R*GPU al-gorithm decomposes a one-way search problem into multiple local search problems which can be solved in parallel. The parallelization way of R*GPU algorithm can be used to solve a variety of heuristic path planning algorithm for parallel speedup.At the last part of this thesis, research works are summarized and many directions for future research works are identified.