Compensating for delays in brain-machine interfaces by decoding intended future movement.

Area of Science:

• Neuroscience
• Robotics
• Biomedical Engineering

Background:

• Brain-machine interfaces (BMIs) typically decode present user intent for prosthetic arm control.
• This leads to system lag due to inherent control delays and arm dynamics.
• Existing BMIs struggle to fully compensate for these delays, impacting user experience.

Purpose of the Study:

• To investigate if decoding future intended movements can mitigate lag in prosthetic arm control.
• To determine if predicting user intent improves brain-machine interface performance.
• To optimize BMI control by aligning with future user commands.

Main Methods:

• Offline analysis of neural decoding accuracy for present versus future movements (up to 200 ms lead time).
• Online testing of a prosthetic arm controlled by a brain-machine interface in a non-human primate.
• Systematic variation of the future prediction time lead in the BMI controller.

Main Results:

• Future movement predictions achieved similar accuracy to present movement predictions.
• Prosthetic arm control performance improved as the future prediction time lead increased.
• Optimal performance was achieved when the prediction lead time matched the system's inherent delay.

Conclusions:

• Decoding future intended movements is a viable strategy to reduce lag in brain-machine interfaces.
• Predictive control significantly enhances prosthetic arm performance and responsiveness.
• This approach offers a promising direction for developing more intuitive and effective neuroprosthetics.