Region Roadmap Connection in Parallel Sampling-Based Motion Planning

In today’s society, parallel computing is commonplace. Motion planning is a computationally complex problem in which solution methodologies can exploit parallelism for complex problems. In one approach to parallel sampling-based motion planning, the environment is divided among processors so that roadmaps can be constructed for each region of the environment independently. Afterwards, the roadmaps in the regions are connected to form the final roadmap. In previous work of connecting roadmaps, the method used was to attempt to connect the largest connected component of the roadmap in one region to the largest connected component of the roadmap in an adjacent region. This method is naive and does not work well for all environments. In this work, connection attempts are made between the largest k1 connected components of the roadmap in one region and the largest k2 connected components of the roadmap in an adjacent region. In this paper, we describe the method in detail and show experimental results explaining the benefits of the approach, namely providing an improved chance of constructing a complete roadmap for complex