Gait Classification-A Study Of AKF's Personal Navigation Method

Micro electro mechanical system(MEMS)inertial sensor has the advantages of small size and high cost performance,so it is often used in portable or small equipment environment.However,the sensors of inertial devices will be affected by some factors,which will accumulate errors in the working process,make the errors larger and larger,and lead to system divergence.To solve this problem,the zero speed correction method is widely used in pedestrian navigation system.The zero speed correction method can eliminate the accumulated error of the device when the pedestrian is at zero speed.However,when the extended Kalman filter is used for data estimation,the parameters of the Kalman filter are fixed,and the accuracy of the Kalman filter is easily reduced when the environment or motion changes.Therefore,in the pedestrian navigation system,the optimal parameters of Kalman filter are determined by adaptive means,which can reduce the positioning error caused by fixed parameters and improve the accuracy of pedestrian navigation positioning.In this paper,an adaptive Kalman filter is designed,which can adaptively adjust the parameters of pedestrian navigation and positioning.The specific process is to first extract the foot motion data features of pedestrians,train the foot features with the random forest algorithm to get the pedestrian motion state,then input the motion state and the corresponding foot motion features into the fuzzy inference system,infer and output the parameter control coefficient of the Kalman filter,and finally adjust the parameters of the Kalman filter with this control coefficient to realize the adaptive Kalman filter wave filtering.This system can reduce the interference caused by the change of motion state,track the track of pedestrian accurately and get the optimal positioning output.The results show that the positioning error of the whole navigation system is less than 2 meters per 100 meters when the pedestrian moves in multiple states,which realizes highprecision pedestrian navigation and positioning in multiple states.This method can be used in hospitals,shopping malls and other occasions.