Implementation of Multi-valued Fuzzy Behavior Control for Robot Navigation in Cluttered Environments

Navigation in environments that are densely cluttered with obstacles is still a challenge for Autonomous Ground Vehicles (AGVs), especially when the configuration of obstacles is not known a priori. Reactive local navigation schemes that tightly couple the robot actions to the sensor information have proved to be effective in these environments, and because of the environmental uncertainties, fuzzy behavior systems have been proposed. The most difficult problem in applying these systems is that of arbitrating or fusing the reactions of the individual behaviors. This paper presents an architectural design of a fuzzy behavior based system for navigation control of robotic vehicles using multivalued reactive fuzzy behaviors. This design allows the robot to thoroughly use the available sensor information when choosing the control action to be taken. Experimental results show that the proposed method can smoothly and effectively guide a robot through cluttered environments such as dense forests.