Obstacle guided RRT path planner with region classification for changing environments

The Rapidly-exploring Random Tree (RRT) has been widely used to solve path planning problems and well suited to lots of problem domains for its probabilistically complete. However, it is not so rapid in changing environments, troubled with moving obstacles and difficult regions. In this paper, a variant of RRT is proposed which is called obstacle guided RRT (OG-RRT), aiming to plan a path in changing environments efficiently. By preserving a group of invalid configurations blocked by obstacles, an entropy value is introduced to label every state in the tree with region classification information. Then a differentiation strategy is adopted to the framework for extending. Finally, with recording the change between invalid and valid nodes, a fuzzy estimation for obstacles´ movements and an opportunistic strategy for reusing information from previous queries will be used to replan a solution fast. In plentiful experiments, OG-RRT is very effective in changing environment.