A body-machine interface for the control of a 2D cursor

摘要: This paper develops a body-machine interface for the control of powered wheelchair using upper-body motion. Our goal was to infer cursor's kinematics from signals recorded 4 Inertial Measurement Units placed on subject's shoulders. We specified Kalman filter measurement model that assumes Euler angles, angular velocities, and linear accelerations shoulders are stochastic function position, velocity, acceleration virtual cursor. learned system encodes cursor movement along with training data. Experimental results show taking advantage redundancy signal improves performance during center-out reaching task. The resulting algorithm provides platform people high-tetraplegia communicate their intended motor actions environment specialized assistive devices.