Re-planning Using Delaunay Triangulation for Real Time Motion Planning in Complex Dynamic Environments

Motion Planning in an unknown, time-varying environment is a relevant problem in present day robotics. While there exist many probabilistic and resolution complete algorithms for motion planning in dynamic environments, each of these are limited by various constraints. This paper presents triangulation based motion planning in unknown dynamic environments. Our novel idea is implemented in two stages. In the first stage, a global path is computed considering only static obstacles in the configuration space. The second stage of implementation uses our proposed algorithm T-star (T*) for replanning the global path according to real time considerations. We demonstrate the effectiveness of our novel idea in multiple simulation environments with more then 20 simultaneously moving obstacles.