A smooth position-force controller for asbestos removal manipulator

Abstract

In robotic grinding tasks, the robot controller must reactively adapt to sudden changes in the environment and be able to handle uncertainties in texture, change in materials, and disturbances caused by vibrations, impacts and friction of the grinding tool operating at high rotational speed. This requires managing sudden signal changes in sensor data. In this paper, we present a smooth hybrid force/position controller based on distance measurements from radars, enabling the robot to achieve a stable interaction with the environment while grinding an unknown three-dimensional surface. The control uses an actively compliant wrist that maintains a desired force centered on the disc and normal to the surface. Our controller is based on a smooth transition between free space and contact modes, significantly reducing the impact force. Additionally, the vibro-dynamic effects are suppressed and smooth environmental tracking is ensured by the impedance/admittance control of the wrist. This framework is validated on a 6-dof anthropomorphic arm through dynamic simulation. The controller is able to adapt reactively to abrupt disturbances in the environment (ex: sudden impacts on the disc) while ensuring good position and force tracking performance.