An Active Control Method for a Lower Limb Rehabilitation Robot with Human Motion Intention Recognition.

Area of Science:

• Robotics
• Biomedical Engineering
• Machine Learning

Background:

• Lower extremity exoskeleton rehabilitation robots (LEERRs) require sophisticated control for effective human-robot collaboration.
• Accurate recognition of human motion intention is crucial for seamless exoskeleton assistance.
• Existing control methods often struggle with the complex dynamics of wearer-exoskeleton systems.

Purpose of the Study:

• To develop an active control method for LEERRs that integrates human motion intention recognition.
• To improve body weight support and center of mass compensation using a vision-driven strategy.
• To enhance gait tracking performance and real-time motion estimation.

Main Methods:

• A vision-driven follow-and-track control strategy for body weight support.
• A muscle-machine interface using a bi-directional long short-term memory (BiLSTM) network to decode surface electromyography (sEMG) signals.
• Optimization of BiLSTM hyperparameters with quantum-behaved particle swarm optimization (QPSO) for a QPSO-BiLSTM model.
• A dual radial basis function neural network adaptive sliding mode controller (DRBFNNASMC) for precise torque generation.

Main Results:

• The vision-driven system accurately tracked human motion trajectories.
• The QPSO-BiLSTM model demonstrated superior prediction of continuous lower limb motion compared to traditional methods.
• The DRBFNNASMC controller achieved better gait tracking performance than FCASMC and PID controllers.

Conclusions:

• The proposed method enables effective active control of LEERRs through accurate human motion intention recognition.
• The QPSO-BiLSTM and DRBFNNASMC provide a robust framework for advanced exoskeleton control.
• This approach holds significant potential for improving lower extremity rehabilitation outcomes.