Control of a Robot Based on Fusion of Visual and Force/Torque Sensor Information. Manipulation of a Deformable Object.

Abstract

Since there exist a lot of deformable objects in our daily life, method of manipulation of a deformable object for a robot is expected not only in industrial manufacturing but also in life support. Generally, a force/torque sensor is used in playing a contact task for a rigid object by a robot. However, only a force/torque sensor is not enough in handling of a deformable object by a robot. In this paper, a fusion algorithm by integrating data from both a force/torque sensor and a visual sensor is proposed for a robot to play a contact task in which an aluminum peg fixed on the robot hand is inserted into a hole installed on a deformable plate. To complete the task, a high-speed camera is used and a real time processing algorithm is developed to detect deformation of the central line of the hole. Then, the results of visual processing are integrated with data of the force/torque sensor using the proposed compliance control algorithm for a robot in playing the task. The effectiveness of the proposed method has been demonstrated through experiments.