Sensor-less external force detection for industrial manipulators to facilitate physical human-robot interaction

Sensor-less external force detection is important for industrial robots which are usually not equipped with external force sensors to be applied in physical human-robot interaction (pHRI). This paper adopts the dynamic models of the robot in both dynamic mode and quasistatic mode to detect the external force. In the dynamic mode, the inertia and friction parameters of the robot are identified with the weighted least squares. The excitation trajectory for parameter identification is optimised. The un-modelled peak points in the joint torque residual are removed by a statistical method. The torque changes of joints in quasi-static mode which are equivalent to the joint pre-sliding friction is modelled with a lumped parameter model, generalised Maxwell slip (GMS) element model. Therefore, there is no need for the switching between the friction models in different modes and this and facilitates the application of dynamic model in the external force detection. The dynamic models of robots both in dynamic mode and quasi-static mode and their effectiveness for external force detection in pHRI are verified by experimental results.