Sensitive Collision Detection of Second-Order Generalized Momentum Flexible Cooperative Joints Based on Dynamic Feedforward Control

Abstract

In this study, a dynamic feedforward second-order generalized momentum flexible cooperative joint collision detection algorithm is proposed, in order to achieve more sensitive and rapid collision detection, in tasks involving human-robot interaction. A second-order disturbance torque observation algorithm is built, based on the dynamic model of a flexible cooperative joint,while the feedforward optimization factor is introduced, to improve the detection accuracy and real-time performance. Next, the interactive external torque is decoupled and the Butterworth optimal filtering process is performed on disturbance observation data, as derived by the respective algorithm, which was verified by simulation. Finally, the proposed generalized momentum external force observer is verified in various dynamic environments, using a self-developed flexible collaborative joint platform. Experiments show that, the proposed observer is very sensitive to the disturbance fluctuations of the external force and can reliably follow these changes, while eliminating the influence of noise caused by friction force. Under different stiffness conditions and artificial disturbances, the proposed approach can break through the low set threshold, in order to achieve fast collision response. The algorithm described in this paper is developed based on a single-axis flexible joint model, which can be quickly applied to any multi-axis collaborative robot, with or without torque sensors.