LAUSR

Abstract According to previous studies, usually multi-degree-of-freedom (DOF) manipulators are installed on robots to perform the valve turning task, and vision systems are used to detect the valve data. However, vision systems are limited in the environment with less visibility and the internal force always occurs as the multi-DOF manipulator is compelled to follow the valve exactly by a firm grasp. In this paper, a novel valve turning method based on force sensing is proposed and implemented to the six-legged robot. Firstly, the method can enable the robot to judge the valve position through force sensing in the case of poor vision situation. Moreover, a 0-DOF tool is installed on the body of the robot. By integrating with the body, the tool has enough workspace to perform the operation task. Compared with the multi-DOF manipulator, loose grasp achieved by the 0-DOF tool helps to reduce the internal force during the valve turning process. In addition, the real-time trajectory planning strategy is designed for the robot. Based on the above initiatives, experiments are carried out to demonstrate effectiveness of the method.