Towards a naturalistic brain-machine interface: hybrid torque and position control allows generalization to novel...

Area of Science:

• Neuroscience
• Robotics
• Biomedical Engineering

Background:

• Brain-machine interfaces (BMIs) enable control of devices using neural signals.
• Current BMIs primarily rely on kinematic control (position/velocity) for tasks.
• Natural movements require force and torque control, not just kinematics.

Purpose of the Study:

• To demonstrate real-time neural control of torques for a virtual arm in non-human primates.
• To investigate the impact of torque control versus position control on movement naturalness.
• To explore the efficacy of torque control in novel dynamic environments.

Main Methods:

• Utilized real-time neural control of torques to operate a virtual arm in non-human primates (M. radiata).
• Compared pure torque control with hybrid torque-position control strategies.
• Recorded neural signals from contralateral or ipsilateral M1 (primary motor cortex).

Main Results:

• Neural control of torques resulted in more natural, ballistic movements compared to position control alone.
• Hybrid torque-position control altered feedforward BMI movement behavior.
• Control was effective using neural recordings from either M1 side.
• Hybrid control adapted to novel external dynamics, mimicking natural movement adjustments.

Conclusions:

• Direct torque control offers a more intuitive way to execute movements with neuroprosthetic devices.
• This approach is particularly beneficial for handling dynamic and changing environments.
• Future sensorimotor neuroprostheses can benefit from direct force/torque control, reducing reliance on kinematic decoders.