An Intelligent Approach for a Two-robot Team to Perform a Cooperative Task

Abstract

Complex future space missions will increasingly rely on human-robot or robot-robot collaboration, which requires the robots to have intelligent capabilities. This paper introduces a decentralized approach of two independent robots to cooperatively accomplish a common task. The task of the robot team is to transport a payload to outside a room through a narrow door. There is no pre-programmed motion trajectory for the two robots to follow and no communication between the robots which means each robot does not know how the other robot will react at any instant. The only information available to each robot is its own position and the pose of the payload. Genetic Fuzzy System (GFS) is employed to train the controllers of individual robots. The kinematics and dynamics models are considered and simulated in MATLAB environment, which have shown that the proposed intelligent control method can be trained to control the two individual robots to successfully achieve their common goal. To demonstrate the effectiveness of this approach, the robot team is trained with a set of scenarios and then verified with an extensive test set.