Research On Human Motion And Position Tracking Method Based On Wearable Sensors In Human-Robot Collaboration

With the development of the robotics industry,human-robot collaboration has attracted more and more attention.Human-robot collaboration can combine the characteristics of high flexibility and creativity of workers with the advantages of high intensity and high precision of robots,but the safety risks of operators in a vulnerable position are relatively large if collisions occur when people and robots work together.Therefore,the research on human position and posture tracking method can effectively improve the efficiency and safety of human-robot collaboration by tracking the posture and position of the human body,determining the distance between the human and the robot in real time,and avoiding the occurrence of collision.In this paper,the human position and posture tracking method is studied,including three parts:data fusion algorithm,human kinematic analysis and software simulation implementation,and the main research contents and the obtained results are shown as follows.First,this paper introduces the current research status of human position and posture tracking methods at home and abroad,analyzes the advantages and disadvantages of several common human pose tracking methods,and determines the use of wearable sensors with high accuracy,low cost and large tracking range for human position pose tracking.The attitude estimation of the wearable sensor is obtained by fusing the acceleration data,angular velocity data and magnetic field data output from the wearable sensor through the extended Kalman filter fusion algorithm.Then,by analyzing the characteristics of human bones and joints and simplifying the real human bones,a 3D human simulation model that can accurately express the human motion posture is established using SolidWorks.The kinematic base point is located at the ankle.On the one hand,the model uses the leg posture transformation to calculate the human position,avoiding the accumulated error caused by the acceleration integral,and on the other hand,it obtains the complete human posture according to the characteristics of the skeletal cascade.However,the foot-root kinematic model is not applicable to the case of simultaneous motion of both feet.and the continuous human tracking cannot be completed by using this model alone.Therefore,a pelvic kinematic model with the base point located at the center of the pelvis is established.The continuous position and posture tracking is completed by real-time dynamic switching between the two kinds models.Finally,the human position and posture tracking software is designed and developed in Visual Studio 2019 using C#development language,and the real-time tracking and motion reproduction of human position and posture is realized by integrated button and visualization interface.In addition,the accuracy and stability of the human position and posture tracking system based on wearable inertial sensors are verified,in which the position tracking accuracy is compared with the actual measurement value and the posture tracking accuracy is compared with the tracking results of the Kinect-based human posture tracking system.In summary,this paper develops a wearable sensor-based human position posture tracking system,which circumvents the problem of line of sight occlusion during human motion tracking,in which the wearable sensor can be bound to any position of the human torso in any posture.The system is characterized by high tracking accuracy and stability,which lays a theoretical foundation and technical support for future applications in the field of human-computer collaboration.