Abstract:

This paper presents multi-sampling strategy approach for Rapidly exploring Random Tree path planning algorithm. The pseudocode is provided and each sampling strategy is explained in detail. Different strategies are used in the bi-tree implementation of RRT based on weights that describe the probability of the related sampling strategy being chosen. The genetic algorithm is used to optimize these parameters on a number of prepared testing boards. The results of the optimization experiment are presented and analyzed, showing strong tendency towards an unstable solution creating a statistically shortest path despite the introduction of a penalty function.