A New Fuzzy-Based Spatial Model for Robot Navigation among Dynamic Obstacles

This paper presents a new approach based on artificial potential fields (APF) method which provides simple and effective motion planners for practical path planning in fully dynamic environments. We have exploited the fuzzy modeling to define fuzzy artificial potential fields (FAPF) which provides a real-time and flexible path planning, in contrast with classical approaches which make use of analytic harmonic functions to solve the navigation problem. It is shown that FAPF paves a way to merge both global and local path planning strategies. In this particular application, the proposed fuzzy system is used to obtain the gradient of the harmonic functions via weighting the membership functions of obstacles and goal. Our proposed technique 1) constructs a "good" fuzzy model for APF; 2) efficiently escapes local minima occurred in path planning using virtual obstacle concept; and 3) can combine the advantages of both the local and global approaches. A path planner based on this technique has been implemented. Simulations show that the planner is both very fast and capable of handling the local minima which can trap mobile robots before reaching the goal. Variety of problems which are far beyond the capabilities of previously developed planners is solved.