Research And Realization On Heuristic And Adaptive Hybrid Floor Positioning

With the widespread development of location-based services,the demand for accurate indoor positioning is getting more and more urgent.Floor positioning,as a prerequisite for indoor positioning in multi-story buildings,is particularly important.Though lots of work has been done on floor positioning,because of the feature differences of multiple scenes,single floor positioning cannot meet the application requirements with low accuracy and robustness in complex multi-story buildings with large hollow areas.Therefore,this paper focus on multi-sensor hybrid floor positioning.To obtain accurate and robust floor estimation in complex multi-story buildings,taking advantage of the natural complementarity of Wi-Fi/barometer while floor positioning,we propose two hybrid floor positioning methods based on machine learning:1.Heuristic hybrid floor positioning method based on XGBoost.This method utilizes confidence threshold to heuristically integrate Wi-Fi based floor positioning and pressure based floor positioning.Furthermore,floor switch detection and HMM are combined to achieve more flexible,accurate and robust floor positioning.Extensive experiments show that using our proposed method can achieve 99.2%accuracy.In addition,this method can also provide continuous vertical coordinates,which is essential for complete vertical positioning.2.Adaptive hybrid floor positioning method based on residual factor graph.In contrast,this method groundbreakingly propose to adaptively combines Wi-Fi and pressure measurements using a novel residual factor graph.Firstly,we propose to cluster the entire building into several autonomous blocks based on RSSI similarity and spatial proximity.Secondly,we build a fine-grained floor model within each autonomous block to optimize the Wi-Fi floor positioning model in complex environments.At the same time,the relationship between barometric pressure change and height change is modeled.Experimental results confirm that our proposal exhibits remarkable improvement in accuracy,robustness,and heterogeneous device adaptability.Experiments show that the accuracy of the two algorithms is better than the existing floor positioning algorithm.The adaptive hybrid positioning algorithm based on the residual factor graph has higher robustness than the first heuristic hybrid algorithm in more scenarios(different times and different devices).