Detecting Reach to Grasp Activities using Motion and Muscle Activation Data.

Abstract

The performance of activities of daily living (ADLs) is directly related to recovery of motor function after an incident such as stroke. Because the recovery process occurs primarily in the home, many efforts have sought to capture gross body motion and limb motion using wearable sensors. One component of function not easily quantified but nonetheless important is the ability to interact with the environment using the upper extremities. In particular, environmental interaction requires the performance of reach-to-grasp (RTG) tasks. The goal of the proposed approach is to determine the extent to which the commercial Myo armband sensor provides a noninvasive mechanism for monitoring and recording RTG task performance. Our results indicated that accelerometer and rate gyroscope data varied significantly between task types, and that a classifier using motion and muscle activation data was capable of distinguishing between gestures with 93% accuracy.