Performance Evaluation of Force Control and Reaction Force Estimation in Force Sensorless Hybrid Control for Workspace Based Controller

This paper presents the performance evaluation of force control and reaction force estimation in force sensorless workspace-based hybrid control for multi-degrees-of freedom (MDOF) robots. There are two reaction force estimation methods. One is reaction force observer (RFOB) in the workspace, and the other is reaction torque observer (RTOB) in joint coordinate space with coordinate transformation. Considering the characteristics of the machine dynamics, the workspacebased controller uses the equivalent mass matrices. As RFOB and RTOB use these matrices, the estimation accuracy depends on the accuracy of these matrices. Conventionally, workspacebased control characteristics in MODF control systems have been considered based on one-degree-of-freedom control systems or human-in-the-loop teleoperation systems. In this paper, equations and experimental results show how the setting of the equivalent mass matrix affects the interference and estimation error in MDOF hybrid control systems focusing on rubbing motion for polishing or wiping.