Co-Adaptive Brain-Machine Interface Prototype for Continuous Movement Control

Co-adaptive Brain-Machine Interface for continuous movement control enabled by state transition and reinforcement learning.

• A continuous movement control approach combined with the reinforcement learning technique to accurately translate subjects’ movement intention from their brain signals.
• A state transition function is utilized to enable the continuous movement control.
• A non-linear reinforcement learning model is used to adaptively establish the neural-action mapping during usage, which enables the model to adaptively follow the ever-changing brain signals of the subject.

• Enable better performance on the continuous brain control task and maintain stable performance over days
• More successful trials
• Smaller response time
• Smaller inter-trial time

• NeuroRobotic control – reaching and grasping, walking and kicking
• Human Computer Interface- cursor/ Joystick moving and clicking
• VR, AR – telecommunication, teleoperation
• Aging city – elderly home care and design