Adaptive Nonlinear Model Predictive Control for Soft Robotics Applications

Abstract

Computer-based adaptive control technology has made significant progress in the real-time management of complex nonlinear system dynamics and the inevitable uncertainties in soft robots. This paper proposes a method for achieving precise and reliable control of soft robotic systems Through the use of adaptive nonlinear model predictive control schemes. Develop a continuous-time dynamic model to identify changes in the core deformation drive state of a multi-link flexible arm; integrate online parameter estimation and model predictive control algorithms to counteract dynamic disturbances caused by environmental or model changes. Adaptive estimation continuously updates its own parameters based on sensor data to achieve an accurate predictive control system under various conditions or uncertainties. The experimental setup uses numerical simulations and a pneumatically driven soft robotic arm to validate the effectiveness of the proposed controller. Compared to traditional MPC and PID control methods, adaptive NMPC can improve trajectory tracking accuracy, anti-interference performance, and execution safety levels. This feature achieves high-reliability real-time computing performance and is suitable for embedded hardware devices. This study proposes a comprehensive approach to ensure the robustness, accuracy, and efficiency of soft robotic systems in resilient control under uncertain environments.