Estimation of the External Forces on a Robot Hand with Fingertip Tactile Sensors

In human-robot collaboration, physical hand-over-hand communication of the human operator’s intention is essential for realizing direct and intuitive cooperation. When an external force acts on a robot hand holding an object, the torque exerted on the finger motor becomes a combination of the grasping torque required to maintain the grasping pose and reaction torque to the external force. To analyze intended interaction, we need to recognize multiple contacts inside and outside the fingertips. In this study, an external force estimation method was developed that uses tactile sensors to measure both grasping and external forces acting on the fingertips of a robot hand. Because a tactile sensor cannot measure all the external force components, it is impossible to directly compute the magnitude and contact angle of an external force. The shear and frictional forces lost during tactile sensing are backwardly estimated according to the difference between the estimated and measured torques. The magnitude and contact angle are computed to estimate external force, which can be used to determine the intention of the user. Simulations based on real sensor responses and experiments were performed to validate the proposed estimation method.