Research On Step Length Estimation Algorithm In Indoor Positioning System Based On Inertial Sensor

Indoor positioning system based on inertial navigation has many advantages,such as low cost,easy to use,working without external sources,so it has a strong adaptability.With the popularity of smart phones,the positioning system based on inertial sensing equipment installed in the upper body(also known as the handheld indoor positioning system)has become a research hotspot.Due to the diversity of human gait,so far,there are lack of a high-precision,common step length estimation algorithm for the different scenarios in the handheld indoor positioning system.The paper aims to propose an adaptive step length estimation algorithm to realize the accurate estimation of the step length in the handheld indoor positioning system.The paper analyzes the waist motion characteristics during pedestrian walking,designs the step length estimation algorithm based on the "linear inverted pendulum"model,and analyzes the influencing factors of the regression coefficients in the model and step length estimation error.Applies the BP neural network to estimate the regression coefficients in the "linear inverted pendulum" model.In order to estimate the regression coefficients accurately,the neural network structure design and input feature set selection are considered.A strategy called "fast selection interval,skilled node" is designed for selecting the hidden layers number and the nodes number.A method combined ReliefF and forward sequence selection(SFS)is used for input feature set selection.Finally,a step-size estimation system based on neural network-based adaptive motion model is designed for indoor positioning applications based on inertial sensors.Experiments show that the step length estimation algorithm based on the neural network-based adaptive motion model can effectively estimate' the distance in the indoor positioning system,the average error of the algorithm is 5.23%.The adaptive step length estimation algorithm proposed in this paper has theoretical and practical significance in handheld indoor location system for estimating distance,and it has good application prospects in emergency response,medical care and location-based service.