Research On Vehicle Indoor And Outdoor Seamless Positioning System Based On Multi- Sensor Fusion

With the growth of demand for location-based services,in the indoor and outdoor environment,continuous seamless positioning and location services has become an important part of the smart city,the smart earth,seamless navigation system has attracted more and more attention.Seamless positioning and navigation system uses the integrated sensors to provide the location,speed,attitude and other information like real-time monitoring of personnel,vehicle position,so as to provide personnel,vehicles and other location-based services.According to the demand of seamless positioning and putting in the industrial warehouse as the background,this paper proposes a vehicle seamless positioning system based on multi sensor fusion,the main contents include indoor positioning based on machine learning,vehicle seamless positioning based on interactive multiple model particle filter,and a vehicle indoor and outdoor seamless positioning platform.First of all,in order to solve the problem of the satellite signal loss of action in the indoor environment,and bad WiFi indoor positioning accuracy,this paper uses the iBeacon wireless sensor network as the basic facilities for indoor positioning,and proposes a machine learning indoor positioning method,first using neighborhood averaging filter to denoise RSSI preprocessing of fingerprint database,then aiming at the characteristics of discontinuous corresponding establishment of the Radio Map and the real physical environment,this paper presents a calculation method that is the double clustering genetic localization based on SVR learning.According to the characteristics of iBeacon signal intensity distribution,first uses clustering algorithm to partition the sample area,and then uses the genetic algorithm and twin support vector regression algorithm for more accurate positioning in each sub region.Experiments show that the method can effectively provide the accuracy of indoor positioning system,and reduce the computational time while using complex machine learning algorithm.Secondly,due to big difference between indoor and outdoor environments,there hardly exists a general localization algorithm for both situations.This paper presents a multiple model based localization algorithm both suitable for indoor and outdoor application.In this work,a switched scheme is proposed to deal with this problem.Because of nonlinearity in measurement update,particle filtering based method is adopted here as the mode matching filter.In time update,movement between two consecutive frames is provided by IMU.In measurement update,GPS and iBeacon are used as observations in outdoor and indoor,respectively.A weighted error compression ratio which can adjust Markov parameters online is proposed to deal with mismatch of competition model brought by model switching frequently.Thus the localization precision in situation of frequent switch between indoor and outdoor can be guaranteed.Experimental results demonstrate precision and robustness of the proposed method outperforms some state-of-the-art methods(e.g.,IMM-PF,IMM-KF,single model algorithm,etc),especially when observation noised is uncertain.Finally,to solve the cross docking navigation problem of warehouse logistics vehicles in industrial warehouse environment,this paper designed and developed the Cyberfleet indoor and outdoor seamless positioning system.The system includes intelligent terminal client application and the positioning cloud platform,meets the requirements of high load and high concurrency technology,and has the advantages that are interactive friendly,easy to use features.The formal operating system proves the effectiveness and feasibility of the vehicle seamless positioning method based on multisensor.