K-Means derived strategized placement of starting points for parallel RRT’s

Abstract

Initial placement of RRT plays an important role when it comes down to improving performance in terms of uniform exploration of map. Algorithms like Distributed , K-Distributed for parallelising of RRT have left the question of initial placement of RRT modules unanswered. Strategised placement of RRT modules can yield parametrically faster performance with a more homogeneous exploration as compared to designs where placement is done randomly. For n > m, m RRTs with strategised placement can perform better than n RRTs with non-strategised or random placement. This strategised placement of RRTs is achieved by clustering the map in m clusters using complex implementation of K-Means clustering algorithm which is sensitive to geometrical constraints of the map. With our work, we propose strategised placement of starting points for RRT that can yield faster performance with a more homogeneous exploration.