Data Acquisition And Force Analysis Of Weight-bearing Walking Based On 3D Human Pose Estimation

Weight-bearing walking scenes can be seen everywhere in real life,such as carrying heavy objects,hiking with bags,walking on uneven roads,and so on.Weight-bearing walking may cause certain injuries to the human body,among which the knee joint is the most common injury site in weight-bearing sports,which seriously affects the walking ability and quality of life of the human body.By designing the knee exoskeleton,it can help the human body to reduce the stress on the knee joint and ensure health when walking with weight.The design of such a weight-bearing walking-assisting exoskeleton requires a reasonable force analysis of the moving human body to understand the action of the force,and then determine key indicators such as the operating mode,carrying capacity,and power source of the equipment design.This article is oriented to the scene of human body weight-bearing walking movement,using target detection,residual network and time series feature extraction algorithm,to develop a three-stage 3D(Three-Dimensional)human body pose estimation algorithm,to realize the movement data of human body weightbearing walking through motion video images,and Kinematic analysis of the knee joint.This method can be used to explore the distribution and conduction of external forces in the body when the human body walks with a heavy load on its back.Provide a new theoretical basis for human motion data collection and motion mechanics analysis.In order to achieve this goal,the work completed in this article includes:1.Design a motion capture experiment for the weight-bearing walking motion scene,collect human body weight-bearing walking motion capture data and videos,and form human body weight-bearing walking motion capture including human body motion video data,human body area labeling,and human body two-dimensional and three-dimensional joint point coordinate information data set.Provide the human motion training set and verification data set required for training for the three-segment 3D human pose estimation algorithm.The article first designed a motion acquisition experiment to construct a walking scene with a load,collected human motion image data and three-dimensional motion coordinates,and manually marked the human body position and two-dimensional joint point coordinates.To lay a good data foundation for the subsequent development of 3D human body posture estimation based on collected images and videos,human body motion mechanics analysis,weight-bearing motion data acquisition,and force analysis integrated system work.2.A three-stage weight-bearing movement 3D coordinate estimation algorithm is proposed.This paper proposes a three-segment human weightbearing movement 3D coordinate estimation algorithm based on motion video data for the weight-bearing walking movement scene,and conducts an algorithm experiment verification on the test data set.Using this algorithm,the human body motion video captured by a common monocular camera can be directly used to calculate and obtain the threedimensional joint point motion data of the human body’s weight-bearing walking motion,and provide motion 3D coordinates for further mechanical analysis research and motion auxiliary equipment design.At the same time,the motion 3D coordinates output by the algorithm are filtered and data post-processed,and based on the development of Unity,the visual simulation of the human body motion process with motion 3D coordinates is realized.3.Establish a human body weight-bearing walking motion mechanics model,analyze the human body motion mechanics for the weight-bearing walking motion process,and calculate the knee joint moment of the weight-bearing walking motion.By using the motion 3D coordinates collected by the 3D human body posture estimation algorithm,the kinematics and mechanics of the human body’s weight-bearing walking process are carried out,and the gait cycle of the weight-bearing walking human body is quantitatively described.Kinematic analysis.Then,a mechanical model was constructed based on the knee joint motion data to study the relationship between the knee joint moment and the load-bearing mass in the load-bearing walking scene.And the calculation results are verified by Opensim simulation software.4.With the weight-bearing walking scene as the background,with the three-stage human weight-bearing movement 3D coordinate estimation algorithm and the weight-bearing walking knee joint kinematics model as the core,an integrated system for human weight-bearing walking movement data collection and knee joint force analysis is established.This article designs a human body motion capture experiment scheme for weight-bearing walking,and produces a human body weight-bearing walking motion capture data set.Then a three-stage human weight-bearing movement 3D coordinate estimation algorithm is proposed.The motion mechanics model of human weight-bearing walking is established,and the three-dimensional joint point coordinate acquisition and knee joint mechanical analysis of human weight-bearing walking based on motion video are realized.Based on the above content,an integrated system for data acquisition and force analysis of human weight-bearing movement was developed.