Real-time robot path planning using experience learning from common obstacle patterns

In this paper we investigate the problem of online robot path planning in an environment. Our main hypothesis in this paper is that the path planning times for a robot can be significantly reduced if it can refer to previous maneuvers it used to avoid collisions with common obstacles during earlier missions, and adapt that information to avoid obstacles during its current navigation. To verify this hypothesis, we propose an online path planning algorithm called LearnerRRT. Our algorithm utilizes a pattern matching technique called Sample Consensus Initial Alignment (SAC-IA) in combination with an experience based learning technique to adapt to the current scenario. We have conducted several experiments in simulations to verify the performance of LearnerRRT and compared it with a sampling-based planner Informed RRT. Our results show that LearnerRRT performs much better than Informed RRT in terms of planning time and total time to solve a given navigation task. When navigation times and distances traveled are explicitly compared, LearnerRRT takes slightly more navigation time and distance than Informed RRT.