Monitoring Intake Gestures using Sensor Fusion (Microsoft Kinect and Inertial Sensors) for Smart Hom

Smart home technologies help post-stroke patients complete activities of daily living (ADL) independently, while saving their time, money, and extra effort. The patients are otherwise required to visit rehabilitation clinics for formal care. Toward this goal, we present our approach to spot specific $'/¶s of eating and drinking in a home setting. We fuse inertial and Microsoft Kinect sensors WRPRQLWRUWKHSDWLHQWV� intake gestures including fine cutting, loading food, and maneuvering the food to the mouth. For both sides of the body, we measured (i) position of the wrist, elbow, and shoulder; (ii) angular displacements at the elbow and shoulder joints; and (iii) acceleration of the spoon/fork/cup which are held by the subject. The use of Kinect allows distinguishing between healthy and paralyzed body sides which is a common problem in tele-rehab. The system was tested successfully on healthy subjects; because stroke-patients show slower motion in a shorter range, the system would serve them at least equally well.