Rule-Based Human Motion Tracking for Rehabilitation Exercises: Realtime Assessment, Feedback, and Guidance

Abstract

In this paper, we report the design and implementation of a Kinect-based system for providing automated realtime assessment, feedback and guidance to users who are practicing rehabilitation exercises at home without the supervision of physical therapists. The foundation for the system is a rule-based framework that can be used to assess in realtime the quality and quantity of the exercises performed by the user. We demonstrate the capability of the rule-based framework by showing the detailed rules for three common rehabilitation exercises, including bowling, hip abduction, and sit to stand. To test its usability and accuracy, we have used the system in a human subject study with eight healthy users. The results show that with proper empirical parameters in the rules, the performance of these exercises can be reliably assessed in realtime.

Keywords

• Rule-based human motion assessment
• microsoft Kinect
• human activity recognition
• rehabilitation exercises
• finite state machine