Study On The Characteristics Evaluation Of A Kinect-based Upper Limb Rehabilitation System

The development of robot technology and the clinical rehabilitation medical science combing together provides a good opportunity for rehabilitation robot research. Research on exoskeleton arm rehabilitation robot has achieved universal recognition in the field of rehabilitation. Comparing with the traditional means of rehabilitation, rehabilitation training mode of stroke patients is optimized by rehabilitation robot, and doctors will be free from the heavy manual work. Cooperating with challenging and interesting rehabilitation game and natural human-computer interaction game control, it can effectively strengthen the patients’ willingness to rehabilitate and treatment effects.Therefore, this paper firstly analyzed domestic and foreign research progress of upper limb rehabilitation robot, studied the cause of stroke hemiplegia dyskinesia disease and the principle of rehabilitation. Summarized the upper limb rehabilitation system design objectives, methods and principles. Then introduced our five-degree exoskeleton upper limb rehabilitation robots, which can realize planar shoulder, elbow, wrist and surface movements. The system adopts the bilateral rehabilitation strategy, achieving the rehabilitation of the nervous system. To improve the recovery system, this paper studied the hand movements based on access static gesture recognition, tracking 5 hand gestures for the purpose of providing natural interaction pattern through the combination of the methods. Hand movements tracking based on Kinect skeleton tracking, and using the experiments to determine the best tracking range, and to verify the tracking stability and good robustness. We recognized 5 gestures based on access depth image and color space model, analyzed the gesture segmentation effect in 4 kinds of environments. We did the performance test in random environments with 500 examples respectively, the accuracy of gesture recognition based on Kinect is 93.4%, which can meet the needs of rehabilitation game control. Finally, we designed and achieved the "attachment" recovery puzzle game based on Kinect motion tracking. We integrated the whole system for experiment, and verified its feasibility.