Adaptive multichannel FES neuroprosthesis with learning control and automatic gait assessment

Area of Science:

• Biomedical Engineering
• Neurorehabilitation
• Robotics

Background:

• Functional Electrical Stimulation (FES) neuroprostheses aid stroke and Spinal Cord Injury (SCI) patients in gait rehabilitation.
• Previous systems required manual tuning and lacked individual adaptation.
• Recent advancements include learning controllers for adaptive stimulation patterns.

Purpose of the Study:

• To develop a learning full-leg neuroprosthesis for adaptive gait support in SCI individuals.
• To control knee and ankle joint movements through antagonistic muscle stimulation.
• To enable continuous gait phase assessment and personalized rehabilitation.

Main Methods:

• Established a continuous assessment method for knee and foot joint angles during gait.
• Developed a novel Iterative Learning Controller (ILC) independent of step duration.
• Utilized physiological joint angle reference bands for gait pattern warping and comparison.

Main Results:

• The neuroprosthesis successfully generated individually fitted stimulation patterns for 3 out of 4 incomplete SCI participants.
• Two participants showed slight improvements in average foot angles (up to 4°).
• One participant exhibited slight improvements in average knee angles (up to 4°).

Conclusions:

• The proposed method allows direct adaptation to individual gait patterns.
• Preliminary tests confirm the neuroprosthesis's capability to generate personalized stimulation.
• Future work will address sensitivity to knee angle resets, gait fluctuations, and ILC gain tuning.