Cursor control by Kalman filter with a non-invasive body-machine interface.

Abstract

We describe a novel human-machine interface for the control of a two-dimensional (2D) computer cursor using four inertial measurement units (IMUs) placed on the user's upper-body. A calibration paradigm where human subjects follow a cursor with their body as if they were controlling it with their shoulders generates a map between shoulder motions and cursor kinematics. This map is used in a Kalman filter to estimate the desired cursor coordinates from upper-body motions. We compared cursor control performance in a centre-out reaching task performed by subjects using different amounts of information from the IMUs to control the 2D cursor. Our results indicate that taking advantage of the redundancy of the signals from the IMUs improved overall performance. Our work also demonstrates the potential of non-invasive IMU-based body-machine interface systems as an alternative or complement to brain-machine interfaces for accomplishing cursor control in 2D space. The present study may serve as a platform for people with high-tetraplegia to control assistive devices such as powered wheelchairs using a joystick.