Machine Vision-Based Motion Detecting System For Rehabilitation Of Lower Extremity Exoskeleton

With the coming of aging society,the disease rate of stroke becomes higher and the number of patients with walking disability goes bigger,which caused the extended need of the healthcare resource.As one of the most important accessories for rehabilitation,lower extremity exoskeleton(LEE)gives a new method for rebuilding walking ability of lower extremity.Wearable exoskeleton has been the main research domain of robotics,whose designs are various,which need a motion detecting system to evaluate the validity of designed exoskeleton.Firstly,this paper analyzed the kinematic model of lower extremity(LE)and introduced the physiological structure and motion feature of joints in LE.Based on the model,This paper described the gait of LE,and fitted the analytic expression of angles for hip and knee joint.Building D-H model for LE,this paper derived the angle formulas for these two joints.Secondly,on the basis of exoskeleton research,this paper redesigned the control system of the LEE to suit for the new trend of domestication.Can-Based kinetic control procedure was programmed in the embedded Linux and ARM processor.With the control procedure,the LEE could assist the patient to walk freely on the ground.The control system reserved network and USB interfaces to realize the communication between patients and doctors.Thirdly,this paper designed a vision-based motion detecting system to evaluate the validity of designed LEE with the standard of human-machine motion consistency.The motion detecting system collected scene data by Kinect and adopted the mark-based method of vision detect.Arranging mark bars on the LE,the system utilized DBSCAN and least square method to setup the D-H coordination system,and calculated the joint angle driven by LEE.Finally,this paper designed an accuracy testing experiment to verify the detecting range and static and dynamic precision of the motion detecting system,which demonstrated the system was suitable for motion capture of the LE.Additionally,the experiment of motion consistency was designed to detect the human-machine coupling characteristics quantitatively.Results showed the designed LEE had the ability to assist the patient to walk.