Neuroadaptive Admittance Control for Human-Robot Interaction With Human Motion Intention Estimation and Output Error...

Area of Science:

• Robotics
• Control Systems
• Biomedical Engineering

Background:

• Human-robot interaction (HRI) is vital for advancing robotics.
• Improving HRI efficiency, robustness, and applicability requires faster response, higher accuracy, and reduced human effort.

Purpose of the Study:

• To develop a novel neuroadaptive admittance control system for natural and stable human-robot interaction.
• To enhance robot performance by accurately estimating human motion intention (HMI) and adapting compliance.

Main Methods:

• Utilized robot interaction force to predict HMI.
• Dynamically updated admittance model stiffness using surface electromyography (sEMG) signals.
• Implemented prescribed performance control (PPC) with an adaptive neural network (NN) for error constraint and uncertainty compensation.

Main Results:

• Achieved human-like compliance through dynamic stiffness adaptation.
• Ensured trajectory error convergence within a predefined range using PPC.
• Demonstrated global uniform ultimate boundedness of the closed-loop system via Lyapunov stability analysis.
• Validated the framework's effectiveness through real-world robot experiments.

Conclusions:

• The proposed neuroadaptive control framework enables natural, stable, and accurate HRI.
• The integration of HMI estimation, sEMG-based compliance, and PPC significantly improves robot interaction performance.
• The system effectively handles uncertainties in robotics, leading to enhanced tracking accuracy and overall system stability.