A haptic human-robot interface accounting for human parameter stochasticity

Abstract

Force feedback haptic devices require physical contact between the operator and the machine, creating a coupled system where the stiffness changes based on that of the operator's arm. The natural human tendency to increase arm stiffness to stabilize motion increases the overall stiffness and reduces stability. Controllers commonly address this with increased damping, which slows the device and decreases operator efficiency. Previous research designed a system to estimate operator arm stiffness by measuring muscle activity and compensate accordingly, modifying the robot's motion based on operator interactions. This achieved the goal of reducing oscillations and increasing performance, but encountered drawbacks related to the unpredictable way in which humans modulate the dynamic parameters of their arm. Controllers designed to be robust to stochastic variation of system parameters are explored, and their effectiveness is validated experimentally. This could further increase the operator performance and reduce fatigue, which could translate into better efficiency and higher productivity.