Integrity Monitoring Technology Of Multi-Source Fusion Navigation And Positioning System

This paper proposes integrity monitoring technology for multi-source fusion navigation and positioning system.Integrity reflects the stability and safety of the system.With the development of positioning technology,facing various complex positioning scenes indoors and outdoors,only a single positioning technology can no longer meet people's positioning needs.At present,the advantages and disadvantages of multiple positioning technologies are complementary to form multi-source fusion navigation positioning technology,which has become a hot research topic in the field of positioning.At the same time,multi-source fusion navigation positioning technology has also been widely applied in various fields.Among many positioning technologies,visual positioning technology is suitable for indoor and outdoor seamless positioning due to its advantages of high precision,low cost,no influence of signal strength,easy combination with other systems,etc.The global navigation satellite system(GNSS),as the mainstream outdoor positioning technology,can be used as the observation input of fusion positioning system as well as visual positioning.However,the two have a common problem:observation errors will spread to the output of the whole fusion system,directly affecting the stability and security of the fusion positioning system.To solve this problem,this paper mainly proposes integrity monitoring algorithms for vision/INS fusion positioning system and GNSS/vision/INS fusion positioning system,including:(1)Studying the positioning principle of vision/INS fusion positioning system,build the system operation platform,analyze the observation model and residual model,and use them to deduce the formula of detection algorithm.(2)Introducing two anomaly detection algorithms suitable for this system:Random Sample Consensus(RANSAC)and traditional algorithm based on parity space.The problems of the above algorithms are analyzed,and an improvement is proposed based on the traditional parity space algorithm:moving feature point detection algorithm,which is used to detect erroneous observation and identify and eliminate.This paper will verify the feasibility of the moving feature point detection algorithm through experiments,analyze the results,and then compare the three detection algorithms with the error detection rate as an index.The experimental results show that the moving feature point algorithm improves the performance of error detection,and the probability of error detection reaches 91.176%.(3)Based on the vision/INS fusion positioning system,it is combined with GNSS to solve the observation model of GNSS/vision/INS fusion system,and proposes a joint detection algorithm based on parity model,and carries out road test experiments.The experimental results show that the positioning effect of GNSS/vision/INS fusion positioning system combined with joint detection is more stable and coherent,and is not susceptible to multipath and other influences.