Research On Pedestrian Navigation Algorithm Based On Adaptive Kalman Filter

Smart devices have become an indispensable part of people's daily lives with the development and popularization of mobile Internet technology.Location-based services have gradually derived tremendous research and commercial value by virtue of the positioning and navigation capabilities provided by such mobile communication devices.Based on the user's real-time geographic location,more accurate advertising,information search recommendations,and other related basic services can be achieved.GPS can provide high-precision positioning results in outdoor open scenes,but it is not suitable for indoor or areas where the signal is blocked.Wi-Fi indoor positioning technology has become a very popular indoor positioning method due to its low cost and the widespread deployment of Wi-Fi access points.However,Wi-Fi signals have poor stability in complex indoor environments and they have the problem of signal attenuation over time.Mobile devices with built-in MEMS inertial measurement units can provide autonomous solutions for tracking pedestrians in different types of environments.This method has lower overhead and does not depend on the external environment.The data used comes from the sensors embedded in the device.The positioning results of the classical PDR system are relatively robust and reliable,but it is difficult to improve the results due to the reliance on the performance of the pedometer algorithm and the difficulty in optimizing the heading accuracy.Compared with the classic solution based on pedestrian stride and heading,this paper builds a scalable extended Kalman filter error model based on the inertial navigation system.It calibrates the current state of the device by maintaining speed,position,and attitude errors.Classic models such as zero-speed update,gravity update,and step position update are used for observation.In addition,a novel gait observation equation based on a human motion model is used to obtain speed measurements.An adaptive heading estimation algorithm using relatively static magnetic field detection is proposed.The algorithm uses a two-way Kalman filter algorithm to mitigate the effects of geomagnetic fluctuations.First,the optimal smoothing algorithm is used to obtain the historical state value sequence,and then the noise parameters are adaptively adjusted and forward recursive to re-estimate the current attitude state.Unlike the solution based on pedestrian dead reckoning,the error model system constructed in this paper contains more process state information,which means that it is more sensitive and expandable,and the implicit constraints between states are conducive to improving the robustness of the model.The experimental results show that the quasi-static magnetic field detection algorithm is available in various complex scenarios,and the experimental results applied are in line with expectations.In addition,the performance of the proposed heading estimation algorithm and related classical algorithms are compared in experiments,and the results show that the error value produced by the proposed algorithm is smaller and more stable.The results of the location tracking experiments performed in the laboratory and the actual mall also show that the system is superior to the traditional step-based course PDR system in terms of flexibility and accuracy,and the algorithm has practical application value and prospect.