Model Predictive Control for Human-Centred Lower Limb Robotic Assistance
This addresses mobility or balance loss from neural trauma by providing a human-centred control approach for exoskeletons, though it appears incremental as it builds on existing MPC and fuzzy logic methods.
The paper tackled the problem of optimal assist-as-needed control for lower limb robotic exoskeletons by introducing a model predictive control architecture with a fuzzy logic algorithm to identify assistance modes based on human involvement, resulting in quick and appropriate transfers among modes and satisfied assistive performance in hardware experiments with three subjects.
Loss of mobility or balance resulting from neural trauma is a critical consideration in public health. Robotic exoskeletons hold great potential for rehabilitation and assisted movement, yet optimal assist-as-needed (AAN) control remains unresolved given pathological variance among patients. We introduce a model predictive control (MPC) architecture for lower limb exoskeletons centred around a fuzzy logic algorithm (FLA) identifying modes of assistance based on human involvement. Assistance modes are: 1) passive for human relaxed and robot dominant, 2) active-assist for human cooperation with the task, and 3) safety in the case of human resistance to the robot. Human torque is estimated from electromyography (EMG) signals prior to joint motions, enabling advanced prediction of torque by the MPC and selection of assistance mode by the FLA. The controller is demonstrated in hardware with three subjects on a 1-DOF knee exoskeleton tracking a sinusoidal trajectory with human relaxed assistive, and resistive. Experimental results show quick and appropriate transfers among the assistance modes and satisfied assistive performance in each mode. Results illustrate an objective approach to lower limb robotic assistance through on-the-fly transition between modes of movement, providing a new level of human-robot synergy for mobility assist and rehabilitation.