A Closed Loop Approach to Human-Robot Handshake

F. Vigni, D. Prattichizzo, M. Malvezzi, E. Knoop

Master Thesis – Siena, october 2018

Abstract – The following work is focusing on the Human-Robot hand interaction, specifically in the grasping force of the handshake. The handshake event between human beings is a well known task, it can enable a communication between participants as a mixture of physical features like: grip force of the hand, velocity approach, duration of the handshake, oscillation frequency and amplitude of the arm. The hypothesis we want to test in this work, is that in human-human handshake there is a balance between an intrinsic (open loop) and extrinsic (closed loop) contributions. Thus, the force applied by a participant in a handshake results as a trade off between the intrinsic force strength and the force perceived from the partner. The target of this work is to develop an experimental setup in order to test the hypothesis for the human-robot handshake grip force, assuming to shape the robot handshake controller as the closest to the one implemented by humans. A 3D{printed object whose shape and dimension similar to a uman hand palm, is used in order to estimate the human grasping force. Relying on the human tactile response, an open loop experiment is run in order to estimate the robot grasping force. A sensorimotor delay is introduced to imitate the reaction time of Central Nervous System (CNS). Five controllers are presented in this work and a user study is run in order to evaluate aspects like: the handshake quality, the human-likeliness of the handshake and the robot personality. The work is born from a collaboration between SIRSLab (Siena) and Disney Research (Zurich) and the results were submitted to RA-L/ICRA 2019 with title: The Role of Closed-Loop Hand Control in Handshaking Interactions. The innovative idea is presented as a force controller obtained considering the inner behaviour of the robot, kept as a constant in this work, and the force applied from the human participant. This method allows for the robot hand to: define its own dynamics and modulate the applied force in order to reach an agreement with the human.