| The integration of virtual environment,machine learning,cloud platform and other technologies in scenario interactive rehabilitation training and evaluation system based on cloud platform is of great value for research and application in the field of intelligent rehabilitation.Virtual enviroment and human pose estimation technology are applied to enhance interest and vividness in scenario interactive rehabilitation training to inspire paraplegic patients’ enthusiasm for completing training initiatively.Besides,objective and scientific evaluation of training effects based on the trajectory of joints is introduced.Additionally,cloud platform technology is implemented to allow doctors to track the rehabilitation training progress of patients in different areas remotely,conduct data-driven rehabilitation evaluation,and provide timely rehabilitation training guidance through the internet.Therefore,the corresponding research on the key technologies of scenario interactive rehabilitation training and evaluation system based on cloud platform is carried out.Above all,according to the requirements of realizing scenario interactive rehabilitation training,storing rehabilitation data on cloud platform and assessing training effects,the system architecture based on public cloud platform technology is proposed,the software framework is described and the main functions are expounded.Unity3D-based scenario interactive virtual environment of rehabilitation training and evaluation is developed oriented towards patients’ different dysfunctional limb and different rehabilitation stages,while the rehabilitation-oriented human-computer interaction is achieved by controlling the virtual agent using the 3D joints position collected by Kinect.Moreover,the similarity of of angle curves between patient and normal subjects is calculated by DTW algorithm as an assessment index of rehabilitation effects.Some human joints are misidentified or even cannot be inferred when partial occluded by rehabilitation training robot using Kinect’s skeletal tracking algorithm.Therefore,a OpenPose-Kinect-based 3D human pose estimation method is proposed and innovatively implemented in the ROS-based progressive scenario interactive rehabilitation training.The method integrates OpenPose-based 2D joint localization with depth information acquired from Kinect to localize joints in 3D,followed by the Holt double exponential smoothing procedure to smooth and predict the trajectory of 3D joints position obtained.Experimental results show that the OpenPose-Kinect-based 3D human pose estimation method is convenient and can meet the real-time requirement,while patients’ 3D posture data during training can also be stored for the subsequent rehabilitation evaluation.Rehabilitation resources suffer from the dilemma of shortage,uneven distribution and high cost.Based on the needs of carrying out rehabilitation training in community rehabilitation centers at home,a rehabilitation data management cloud platform is proposed with the concept of ‘Internet+Rehabilitation’ and initially implemented on the basis of SSM and Shiro framework.The cloud platform is equipped with the function of collecting,storing,analyzing and displaying the rehabilitation data uploaded by rehabilitation clients,and enable rehabilitation doctors to investigate the rehabilitation progress of patients and provide individualized rehabilitation training guidance for patients.The scenario interactive rehabilitation training and evaluation system based on cloud platform can acquire human joint positions in 3D coordinate space with only one Kinect,motivate patients to complete rehabilitation training initiatively,and provide rehabilitation effects evaluation for doctors.The remote monitoring and guidance based on cloud platform can break the space and time restrictions between patients and doctors,and provide a technical guarantee for the realization of a new mode of rehabilitation,nursing and elderly care. |
