A Fuzzy Logic Path Planning Algorithm Based on Geometric Landmarks and Kinetic Constraints

Abstract

This paper mainly focuses on the path planning of mobile robots in complex two-dimensional terrain. It proposes a fuzzy rule-based path planning algorithm for multiple guide points by changing the spatial point-taking method and combining Dijkstra's algorithm and fuzzy logic algorithm. The planning process of this algorithm divide into three stages. The first stage identifies the edge points of the forbidden area by designing the search space, marks the feasible area widths of the edge points in X and Y directions, and marks their midpoints. The second stage uses Dijkstra's algorithm that does the road map sorting on these marked points and the starting and ending points and takes the lowest cost sequence as the search road map. In the third stage, using a fuzzy logic system to search these road signs one by one until the endpoint area is searched. The simulation results show that this algorithm can solve the complex environment that traditional fuzzy inference algorithms cannot plan. Compared with the graph search algorithm, this algorithm dramatically reduces the planning time and provides more flexible turning angles. This algorithm can better consider the robot's size and the relationship between speed and turning angles while estimating the motion state at each step compared with the sampling algorithm. This algorithm will extend to group path planning and dynamic environment planning in subsequent studies.