Gaussian Belief Space Path Planning for Minimum Sensing Navigation

In this article, we propose a path planning methodology for mobile robot navigating through an obstacle-filled environment to generate reference that is traceable with moderate sensing efforts. The desired characterized as the shortest in Gaussian belief manifold equipped novel information-geometric distance function. function introduce shown be asymmetric quasi-pseudometric and can interpreted minimum information gain required steer belief. An RRT*-based numerical solution algorithm presented solve formulated shortest-path problem. To insight into asymptotic optimality of proposed algorithm, show considered length continuous respect topology total variation. Simulation results demonstrate method effective various navigation scenarios reduce costs, such frequency sensor measurements number sensors must operated simultaneously.