A practical computational technique for mobile robot navigation

This paper addresses a practical motion planning strategy which navigates a rigid robot in a static known environment. Unlike the conventional local approaches called the artificial potential field method, a major effort is first directed toward solving elliptic boundary problems expressed by the equation of heat conduction. According to the temperature behaviors calculated at each grid point, the reasonable direction of heat flow is then searched through the free space and is exploited for robot navigation purposes. A computer simulation based on finite-difference techniques shows an obstacle-avoiding and local minima-free motion with the shortest path between the start and goal.