Particle Filter Based Path Planning with Collaborative SLAM
Introduction
Classic path planning problems involve finding a route to a known, static location. Work at NCR has focused on developing path planners for finding a moving target at an unknown location. A preliminary step in planning paths for mobile robots is generating a map of the environment. Recently, several techniques have been developed for simultaneous localization and mapping (SLAM), however the map is only available to the robot performing SLAM; individual robot agents each produce separate maps. By combining maps while minimizing the required network bandwidth (and allowing for disconnection), the map robustness is increased and unexplored areas can be mapped faster. Once mapping is complete, a particle filter based approach can be used to efficiently plan paths to explore areas where the unknown target may be.
