Sensorless Object Exploration and Stable Contact by Robot Manipulator in Unknown Environment

Effective algorithms for object exploration and manipulation have been widely required for the improvement of the dexterity of robots. Especially, feedback of tactile sensation can be utilized for the precise and versatile manipulation. However, most conventional methods are conservative. This is because they ignore the exploration process and assume that a robot manipulates an object at the tip of its arm. These conditions are not always satisfied, especially in the case that the robot does not have the information about the object and have to explore it firstly. In this paper, we propose an algorithm for sensorless object exploration and stable contact. The algorithm is based on the statics with the extended Jacobian matrix, which relates the joints and an arbitrary point of the robot. The exploration algorithm for the stable contact with unknown environment is also introduced, and represented in a form of a state machine. Experimental results with a robot manipulator are shown for the validation of the proposed algorithm. Furthermore, an advanced experiment with a dual-arm robot is shown and validates the grasping motion with the proposed algorithm.