Research Of Orientation Estimation For Human Motion Based On Multi-sensor

Human motion capture system is an important way to make special effects movies,realize virtual reality systems and other human-computer interactive activities,and the principles of motion attitude measurement is the key technology to realize the motion capture technology.Compared with the motion capture system based on optics,acoustics and electromagnetics,the inertial human motion capture system is a passive capture device,avoiding the tedious auxiliary instruments and harsh restrictions.Therefore,it is of great significance and application value to research on the inertial attitude measurement technology.Firstly,the realization of human motion capture system and the inertial sensing theory are deeply studied in this thesis.Accelerometer,gyroscope and magnetometer are selected as the sensors for capturing motion information,and the data of multi-sensor are processed by information fusion technology.According to the forward kinematics principle,the human motion model is established,and the research program of human body posture measurement is completed.Secondly,aiming at the characteristics of low precision and low cost of sensors,and the different characteristics of accelerometer,gyroscope and magnetometer,the attitude fusion algorithms based on the three kinds of sensors are studied,specifically comprises the complementary filtering and the adaptive hybrid filtering.The static,electromagnetic interference and linear acceleration interference state are analyzed in theory,simulation and experimental verification.The results show that the root mean square errors of the hybrid filtering fusion algorithm in magnetic interference and dynamic environment are both less than 2 degree,and the computation time is increased by 25%.The final test data show that the algorithm is effective for the suppression of the error of the sensors.Then,according to the principle of human dynamics,the tree structure is established,and the binding position of the multi-sensor unit for capturing human motion information is studied.Based on the forward kinematics theory,the model of human joint rotation angle and the position model of human skeleton movement are established.The simulation experiments and the human test experiments are carried out to verify the model of human joint rotation angle.The experimental data show that the root mean square error of the simulation experiments and the human test experiments are less than 1 degrees and 5 degrees.The results show that the model is reasonable,and the experimental verification is rigorous.Next,in order to solve the problem that the root space coordinates of the human body kinematics model is fixed,the kinematics model of human leg is established.Based on the analysis of the characteristics o f foot acceleration,the support leg and swing leg are detected,and the displacement estimation algorithm based on the support leg detection is realized.The accuracy of the root node displacement estimation is verified under three kinds of trajectories,which are straight line,curve and stairs.The experimental results show that the offset rates of displacement estimation in three kinds of walking paths are both within 4%,which can get the position of human.Finally,VS2010 and open graphics library are used to create a console program.The motion data is used to drive the virtual human to reproduce the human motion,which further verifies the feasibility of the proposed algorithm and the effectiveness of the practical application.