A Novel Robust Control Method Combining DNN-Based NMPC Approximation and PI Control: Application to Exoskeleton Squat Movements

2025-09-30

AI Summary

To address the poor real-time performance of nonlinear model predictive control (NMPC) and the insufficient robustness of NMPC–deep neural network (DNN) controllers under unknown disturbances during exoskeleton squatting, this paper proposes a hybrid control strategy integrating NMPC–DNN with PI control. Innovatively, it structurally fuses data-driven NMPC approximations with feedback-based PI control within a human–exoskeleton coupled dynamic framework—a first in the literature. A three-joint (ankle–knee–hip) active exoskeleton model is employed, and the DNN is trained on over 5.3 million samples. Experimental results demonstrate that, under unseen disturbances, the hybrid controller achieves significantly lower tracking error than NMPC–DNN alone, reduces computational latency by 99.93%, and decreases RMS joint torque by 29.7%–41.8%, thereby markedly enhancing both dynamic robustness and execution efficiency.

Technology Category

• Intelligent Robots: State Estimation
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Nonlinear Model Predictive Control (NMPC) is a precise controller, but its heavy computational load often prevents application in robotic systems. Some studies have attempted to approximate NMPC using deep neural networks (NMPC-DNN). However, in the presence of unexpected disturbances or when operating conditions differ from training data, this approach lacks robustness, leading to large tracking errors. To address this issue, for the first time, the NMPC-DNN output is combined with a PI controller (Hybrid NMPC-DNN-PI). The proposed controller is validated by applying it to an exoskeleton robot during squat movement, which has a complex dynamic model and has received limited attention regarding robust nonlinear control design. A human-robot dynamic model with three active joints (ankle, knee, hip) is developed, and more than 5.3 million training samples are used to train the DNN. The results show that, under unseen conditions for the DNN, the tracking error in Hybrid NMPC-DNN-PI is significantly lower compared to NMPC-DNN. Moreover, human joint torques are greatly reduced with the use of the exoskeleton, with RMS values for the studied case reduced by 30.9%, 41.8%, and 29.7% at the ankle, knee, and hip, respectively. In addition, the computational cost of Hybrid NMPC-DNN-PI is 99.93% lower than that of NMPC.

Problem

• Addressing NMPC-DNN robustness issues under disturbances
• Reducing computational cost for real-time robotic control
• Improving exoskeleton tracking performance during squat movements

Innovation

• Combining DNN-based NMPC approximation with PI control
• Applying hybrid controller to exoskeleton squat movements
• Reducing computational cost by over 99.9 percent