INTELLIGENT MOBILE ROBOT PATH PLANNING WITH FUZZY SYSTEM APPROACHES

Abstract

The path planning problem for intelligent mobile robots involves two main subproblems; the representation of obstacles in the task environment and the development of a strategy to determine a collision-free route. In this project, new approaches have been developed to solve these subproblems. The first subproblem was solved using the fuzzy system approach where obstacles were represented by fuzzy half-planes. This approach is able to represent obstacles with imprecise, uncertain, or fuzzy geometries. The other subproblem was overcome through the use of a strategy selector, which chooses the best strategy, among a few, to be used for determination of a collision-free route. This approach is able to draw on the strengths of different strategies under different situations. Based on the fuzzy system approach for world representation, three existing strategies were modified for use in the strategy selector. One is for the static environment case while the other two are for the dynamic environment case. These strategies are improved versions of existing methods. The modified path planning strategies and the strategy selector itself have been successfully implemented in Turbo C on an IBM-compatible 386 computer. Simulation examples were used to illustrate the effectiveness of the proposed approaches. Recommendations on areas to further this project have also been included.