A novel model predictive dynamic adaptive impedance control for robotic force tracking on complex surfaces

The time-varying nature and complexity of environmental stiffness make it difficult to achieve accurate and stable tracking of normal contact force on complex surfaces in robotic belt grinding. The more complex the contact environment, the higher the requirements for robotic perception and self-regulation capabilities in relation to the environment. To address this issue, a novel model predictive dynamic adaptive impedance control algorithm (MPRPO-DAIC) is proposed to improve the accuracy of robotic force tracking in complex contact environments. This control strategy consists of a dynamic adaptive impedance control algorithm integrated with a model predictive control framework. The dynamic adaptive impedance control enhances the robustness of adaptive impedance controller (AIC) by introducing a dynamic observer term. Additionally, the rolling parameter optimization algorithm enables real-time optimization of key parameters within the dynamic observer. Subsequently, through the combination of dynamic adaptive impedance control and model predictive control, the robotic velocity compensation value is calculated by solving a quadratic programming problem. Finally, the velocity compensation value is integrated and sent to the robotic end-effector, enabling stable tracking of contact forces in complex surfaces. The simulation and experimental results of force tracking on complex surfaces validate the superiority of the proposed control strategy. The simulation results show that MPRPO-DAIC exhibits optimal force control performance in complex tracking environments. In the force tracking experiments, MPRPO-DAIC improved the force control accuracy by 37.25% and 44.90% compared to AIC on surfaces with large deformation and high-frequency variation, respectively.