Decoupled model predictive control with SCESO for contact force regulation of a shape-adaptive shield in assistive toileting robots

Abstract

To overcome the limitations of discomfort and physical constraints in existing wearable nursing devices, this paper proposes a non-wearable shape-adaptive shield integrated with real-time pressure sensing for assistive toileting robots. The shield employs a bionic continuum structure with multiple adjustable spine segments and flexible sensors, enabling safe body conformity and contact force monitoring across diverse body types. To address multi-input multi-output coupling that compromise force control stability, an integral-type state feedback decoupling method is developed to mitigate inter-segment coupling. Furthermore, a novel control strategy integrating State-Compensated Extended State Observer (SCESO) and Model Predictive Control (MPC) is proposed to optimize dynamic force tracking under uncertainties. Simulations and experiments demonstrate that the decoupling-SCESO-MPC framework achieves superior tracking accuracy (average root mean square error of 0.017 N across 10 subjects with diverse body types) and robust performance, thereby providing a practical solution for safe, comfortable, and adaptive human–robot interaction in assistive toileting scenarios.