Object handling tasks based on active tactile and slippage sensations in a multi-fingered humanoid robot arm

Abstract

This paper presents a new algorithm for object handling tasks based on active tactile and slippage sensations using a humanoid robot multifingered arm for an object that exists at an arbitrary position. The idea is to enhance real-time object handling tasks based on tactile sensing in humanoid robotics, where grasp, move and release motions are involved. We developed a novel hemisphere-shaped optical three-axis tactile sensor to mount on fingertips of the robot arm. The tactile sensor is capable of defining normal and shearing forces simultaneously. For grasp and release motions, we designed the algorithm based on slippage direction analysis consisting of coordinate transformation of the sensing element for the arm global coordinate. The robot control system uses the analysis results to determine whether an object is in contact with the ground without needing to measure the height of the ground. The algorithm was evaluated in experiments with soft and hard objects, whereby results revealed good performance for the robot fingers in handling an object at an arbitrary position.