摘要：AbstractIn this paper, we develop a data-driven optimization method for determining admittance parameters for haptic assist devices, specifically those controlled by a variable admittance scheme. The proposed approach relies on measurable control and kinematic inputs to the system. Typical haptic controllers are implemented by combining two control loops: an outer-loop admittance controller that sets desired system kinematics and an inner-loop position controller that tracks these desired kinematics. Current robot position controllers are able to achieve very accurate results. Our proposed approach is based on three (3) assumptions: 1) a perfect position control and rigid grasp contact, 2) an equilibrium position at the origin, without loss of generality, and 3) an optimization time window small enough to allow certain system parameters to remain constant. Optimal admittance parameters are derived under these conditions, based on trade-off between position error and control effort criteria. Our approach is tested with simulated data and preliminary experimental data acquired from a single subject in the 1-DoF case. The goal of our research is to improve the stability and performance of physical human-robot interaction (pHRI) systems.
