Wireless Inertial Motion Capture System Based On Attitude And Space Time Trajectory

With the continuous improvement of science and technology,action capture technology is more and more appearing in the daily work and life of the public.It involves a wide range of subjects,and has a broad development prospect and wide application in the fields of medical rehabilitation,animation model making,sports,human-computer control,virtual reality and so on.Thanks to the rapid development of microelectronics technology,the volume of inertial sensor chip is reduced,the cost is reduced and the accuracy is improved.Compared with the action capture system based on optical,mechanical and electromagnetic,the action capture system based on inertial sensor has the advantages of low production cost,convenient use,small interference from external environment and large range of activities.Based on the technology of micro inertial sensor,combined with the principle of inertial navigation and data fusion and computer graphics,a set of real-time wireless action capture system which is not limited by light,easy to wear,less motion limitation and high cost-effective is designed.The system consists of 15 sensor nodes,one computer and one router.The sensor node collects the posture data of the moving limbs and sends it to the upper computer through Wi Fi wireless transmission.The attitude data and the time-space trajectory are calculated in the upper computer,which drives the human model to follow the real human body movement.This paper first designs the overall technical scheme of the action capture system,describes the hardware selection,circuit design,data transmission mode,algorithm structure,sensor node deployment position,and effectively corrects the errors of inertial sensor.Based on the overall framework of human sensor network,the zero speed update algorithm based on indirect Kalman filter is used to solve the human motion trajectory;The gradient descent algorithm is used to solve the motion limb posture by integrating accelerometer,gyroscope and magnetometer data to the sensor nodes deployed in the moving limb,and the attitude information expressed by quaternion is output.Then,the human body posture and foot track are combined to realize the real-time tracking of the human body posture and absolute position in space by using the different moving limb traversal update sequence different from the traditional action capture system.A series of experiments have proved that the method can be used in many motion states.In the closed rectangle walking experiment,the average relative position error is 0.31%,the average relative distance error is 1.564%,and the accuracy is high;The reliability of the attitude algorithm is tracked by the bar shape human model in MATLAB.The error of three experiments in the 90 degree experiment of the small arm lift is that the real human body can be followed basically;The three-dimensional reconstruction method of human body fusion attitude and displacement is carried out in different path experiments and optical contrast experiments.The attitude and waist trajectory are close to the actual path and optical estimation results.Finally,in the vs2012 development environment,the software platform of the system is built in combination with Direct X SDK and MFC framework,which can realize the receiving and saving of inertial data,the loading of human body model,the demonstration of real-time action capture animation,and the test of 3D human body model following the human motion.3D action capture animation can accurately track human motion.