Research On Pedestrian Fusion Positioning Method Based On Optimized Heading Estimation And Multidimensional Particle Filter

Nowadays,Location Based Service(LBS)has very important and widespread applications in military,livelihood and commercial areas.Global Navigation Satellite System can provide satisfying location service in most outdoor scenes,but it is influenced by signal attenuation and multipath effect in indoor environment,and thus cannot provide localization in indoor environment.With the increasing popularity of smartphones with high computational performance and many built-in sensors,it is currently the mainstream direction to utilize smartphones' capabilities of environmental perception for indoor pedestrian positioning.First,this paper has carried out three aspects of research using the self-contained sensors of smartphone focusing on heading estimation,one of core problems of pedestrian dead reckoning.(1)This paper proposes a magnetometer calibration scheme for smartphone in indoor magnetic interference environment to improve the performance of heading estimation in indoor environment.(2)For pedestrians who may have different ways of carrying smartphones,this paper presents an algorithm to estimate the pedestrian direction by using the acceleration frequency characteristics of pedestrian,aiming to eliminating the limitation for pedestrians to use and carry smartphones in dead reckoning algorithm.(3)This paper proposes a method to assist heading estimation by the road network data to further correct heading error.Secondly,in order to use effectively many kinds of data from multi sensor,opportunity signal and map information,this paper proposes multi-dimensional particle filter fusion positioning algorithm.(1)This method combines pedestrian dead reckoning and opportunity signal positioning to achieve the complementary advantages of both.(2)In order to solve the limitation of the way to carry smartphones,this paper uses motion recognition algorithm to realize positioning under many kinds of carrying modes by the optimization of step detection algorithm for the special carrying mode and the offset compensation for pedestrian direction.(3)This paper designs multidimensional particle filter to make estimations of many uncertain states in positioning process including two-dimensional coordinates,heading,step length model parameters,motion labels,lifetime,number and factor,to improve the robustness of filters.In the end,this paper realizes a complete indoor pedestrian positioning system based on the smartphone sensor,which can achieve the universal high-accuracy positioning and has certain practical value.