Model Predictive Variable Impedance Control of Manipulators for Adaptive Precision-Compliance Tradeoff

Abstract

Impedance control (IC) is widely used in contact-rich manipulator tasks since it provides manipulators with both operation precision and contact compliance, and trades them off by impedance parameters. However, fixed impedance parameters limit the applicability of IC in complex tasks during which the focus on the operation precision or the contact compliance is variable, such as physical human–robot collaboration and complex assembly tasks. This article presents model predictive variable impedance control approaches for adaptive precision-compliance tradeoff to satisfy variable task requirements. Specifically, we establish a novel impedance model, which transforms the variable impedance law design problem into a control law design problem, and allows us to consider novel impedance constraints that can determine manipulators' extreme precision and compliance properties. According to whether state constraints are further considered, one-step (OS) and multistep (MS) model predictive control approaches are proposed to solve these transformed control problems, and the corresponding optimization problem of the OS MPC can be established as a QP problem due to the special form of the novel impedance model. A tank-based approach is further proposed to correct the variable impedance laws for the system passivity concern. Various comparative experiments are conducted to validate the effectiveness of the presented approaches.