Trajectory Recovery Based On Interval Forward–Backward Propagation Algorithm Fusing Multi-Source Information

With the continuous expansion of intelligent transportation markets such as the Internet of Vehicles and autonomous driving in recent years,various application services such as lane level positioning,map matching,and vehicle trajectory prediction require a large amount of trajectory data as basic support.Therefore,research on real-time vehicle trajectory recovery technology is on the ascendant.When intermittent abnormalities occur in positioning equipment or users are in dense forests,tunnels,and other indoor situations,the positioning sampling rate measured by the Global Navigation Satellite System(GNSS)is insufficient,resulting in the loss of trajectory information.In this context,this paper innovatively applies the Interval Forward Backward Propagation Algorithm(IFBP)to fuse information such as inertial navigation and map constraints to recover and correct multiple sensor information,and improves the reliability of trajectory recovery by converting various sensor information into interval data.The specific research work is as follows:(1)Design of forward and backward trajectory recovery algorithm based on interval analysis.A dead reckoning algorithm based on the fusion of inertial navigation information and other positioning-related information can improve the accuracy of vehicle real-time trajectory recovery,but its results have an error accumulation effect,and the error in the calculation results will become increasingly large.Although information fusion based on the Kalman filtering algorithm can solve the problem of error accumulation,its measurement conforms to the assumption of Gaussian distribution,greatly limiting its estimation accuracy.The IFBP algorithm proposed in this paper uses interval analysis to fuse multiple sensor information.Through forward and backward propagation,it can effectively fuse GNSS position information and inertial navigation sensors,and the estimated trajectory interval determines the actual trajectory interval containing the vehicle.(2)Research on trajectory recovery for multi-information fusion.Based on the traditional multi-sensors data fusion and positioning integration system,it is usually composed of GNSS information-receiving modules,inertial sensors,odometers,etc.Most of these sensor modules output different data types,so there is a challenge for the compatibility of multiple data types of information fusion.In this paper,after reasonably segmenting multi-information source data,interval computation is used to convert various interval measurements into interval trajectory information.The intersection and union algorithms of set theory are applied to process the estimated interval results from different information,improving the fusion accuracy of multi-source trajectory-related information and improving the fusion efficiency.(3)Establish a platform for integrating forward and backward propagation algorithms of map restriction information.The paper designs and conducts an effective multi-sensor information fusion positioning experiment,collecting multiple sensor information and uploading it to the cloud for real-time data monitoring and storage.The hardware platform is divided into a data acquisition and storage system and a vehicle drive and control system to ensure the smooth and effective conduct of the experiment.In this experiment,map constraint information was creatively set and fused with other map information to verify the proposed algorithm’s certainty,compatibility,and stability.