Micro Inertial Navigation Pedestrian Trajectory Measurement And Smoothing Algorithm Research Based On Improved Gait Detection

Trajectory measurement is a technique that digitizes the human body using sensor technology.In recent years,along with the rise of VR games,science fiction movies,and other industries,higher requirements have been put on the smoothness of the captured trajectory,and the trajectory measurement solution of the MEMS inertial sensor has the advantages of small size,convenient wear,and free from environmental constraints and has a good application prospect.The trajectory measurement technology based on MEMS inertial sensors is generally implemented by using the SINS algorithm combined with the zero-speed correction based on Kalman filtering.Although it can suppress the accumulation of navigation errors to a certain extent,it also has some problems.Inaccurate gait detection leads to trajectory distortion.At the same time,the zero-speed correction is only performed at the zero-speed point of the foot,resulting in abrupt changes in the navigation information when the foot transitions from the oscillation to the quiescent state,thus causing the “sawtooth” phenomenon of the trajectory.The lack of correction of attitude errors by Kalman filter will also deepen this phenomenon,and the presence of a "sawtooth" trajectory will cause serious jitter in the character's computer model when it is used to reproduce the trajectory of people.Therefore,the algorithm for the "sawtooth" trajectory smoothing is important significance.Firstly,this paper analyzes the reasons for the inaccuracy of gait detection based on experimental data,and concludes that the reason for the inaccurate detection is the result of the combined effect of the local fluctuation of the test statistic and the determination of a single fixed amplitude threshold value,Specifically,leakage detection and false detection in the zero-speed interval,At the same time,the analysis points out the inadequacy of the time threshold judgment method based on the amplitude threshold judgment,Furthermore,based on the study of the statistical law of the length of the wobbled interval,the dual-threshold gait detection algorithm based on the Layda criterion is proposed,and the zero-speed accurate detection is finally achieved.Secondly,the formation of the "sawtooth" in the trajectory is closely related to the accuracy of the attitude calculation.Therefore,in view of the insufficient correction of the attitude error in the Kalman filtering process,an attitude correction module is constructed to better suppress the divergence of the attitude error.To a certain extent,the smoothness of the trajectory can be improved.At the same time,aiming at the abrupt change of navigation information,a fixed interval smoothing filter algorithm based on Kalman filter is proposed.Combined with the fixed interval allocation method,we can use the observations on both sides of the swing zone to estimate the navigation error of the foot swing process,and then achieve a smooth transition of the foot's trajectory at zero speed and non-zero speed.Finally,several sets of trajectory comparison experiments were designed in combination with the actual walking data to verify the suppression effect of the improved gait algorithm,posture correction module,and backward smoothing algorithm on the track “sawtooth”.Experiments show that the corresponding algorithm proposed in this paper can effectively suppress the "sawtooth" of the trajectory,and ultimately obtain the trajectory of the ideal smoothness.