Genetic-based adaptive fuzzy controller for robot path planning

In this paper, we propose a new methodology to solve the robot path planning problem. This methodology uses a genetic-based adaptive fuzzy controller to navigate the robot from the starting location to the goal location safely and smoothly. For generating the fuzzy rule sets automatically and tuning the parameters of the membership functions adaptively, an encoded knowledge structure of the fuzzy genetic map (FGM) is presented. To prevent the condition that causes too many binary digits for encoding the membership chromosome, we develop a ring structure to reduce the size of the chromosome. In the selection procedure of the genetic algorithm, we develop fitness functions such as safety factor, smoothness factor and minimum-length factor for different criteria. Two types of environmental maps are tested to verify the robustness of the methodology, where the obstacles of the maps are surrounded with the artificial potential field. Compared with other traditional heuristic methods for robot path planning, the proposed methodology shows robust learning capability and better performance.