A Gait Sub-Phase Switching-Based Active Training Control Strategy and Its Application in a Novel Rehabilitation Robot

Area of Science:

• Biomedical Engineering
• Rehabilitation Robotics
• Artificial Intelligence in Healthcare

Background:

• Gait phase recognition is crucial for effective rehabilitation of balance disorders.
• Traditional rehabilitation robots often lack adaptive control and personalized training capabilities.
• Multi-source heterogeneous motion data offers rich information for accurate gait analysis.

Purpose of the Study:

• To develop a heuristic hybrid deep neural network (DNN) model for accurate gait sub-phase recognition.
• To design and implement a novel motion control strategy for a rehabilitation training robot.
• To enhance patient autonomy and engagement in rehabilitation through an active-passive training approach.

Main Methods:

• Fusion of multi-source heterogeneous motion data for gait analysis.
• Development of a heuristic hybrid deep neural network (DNN) for gait phase recognition.
• Implementation of a variable admittance control strategy integrated with gait recognition for robot control.

Main Results:

• Achieved over 99% accuracy in gait phase recognition.
• Developed a novel rehabilitation training robot with an active-passive training control strategy.
• Demonstrated enhanced patient autonomous movement and engagement during rehabilitation.

Conclusions:

• The proposed DNN model and robot control strategy significantly advance balance disorder rehabilitation.
• The active-passive training approach overcomes limitations of single-mode rehabilitation robots.
• This research provides a foundation for personalized and collaborative rehabilitation robotics.