Abstract
Integrating model predictive control (MPC) with impedance control (IC) has emerged as a powerful framework, combining the physical compliance of IC with the explicit constraint compliance and optimization of MPC. However, the impedance models specified by the user or obtained through optimization may be too conservative to achieve an adequately fast convergence or too aggressive to ensure interaction safety. In this work, a control framework named EMSIC is initially developed. With EMSIC, the controlled system selects between a task-preferred and a conservative impedance model based on safety conditions while complying with MPC constraints. The proposed framework is then modified to achieve predefined-time (PdT) convergence based on a user-defined convergence time, as well as PdT-synchronized convergence for multistate systems. Finally, comprehensive physical robotic interaction experiments are conducted to demonstrate the practical effectiveness and performance of the proposed control framework.