Development of a closed-loop feedback system for real-time control of a high-dimensional Brain Machine Interface

Area of Science:

• Neuroscience
• Biomedical Engineering
• Robotics

Background:

• Advancements in neural prosthetics necessitate efficient testing methods for Brain Machine Interface (BMI) algorithms.
• Current methods for testing BMI decoding algorithms can be time-consuming and expensive, often requiring physical prostheses.

Purpose of the Study:

• To create a virtual prosthetic limb for the real-time testing of complex BMI decoding algorithms.
• To provide a cost- and time-efficient alternative to physical prosthetic development for BMI research.

Main Methods:

• Developed a 27 degrees-of-freedom (DOF) virtual macaque upper limb avatar using x-ray and MRI-guided skeletal reconstruction and artistic rendering.
• Animated the virtual avatar in real-time using kinematic data from awake, behaving macaques captured by a 16-camera motion system.
• Implemented a custom software interface for real-time animation and control.

Main Results:

• Successfully created a functional virtual prosthetic avatar with 27 DOFs.
• Demonstrated real-time, closed-loop control of the virtual prosthetic using kinematic data.
• Validated the system for testing high-complexity BMI decoding algorithms.

Conclusions:

• The developed virtual prosthetic system offers a practical and economical approach for evaluating BMI decoding algorithms.
• This method significantly reduces the costs and time associated with testing advanced neural prosthetic control strategies.
• Facilitates rapid iteration and refinement of BMI decoding algorithms in a simulated environment.