Research On Multi-sensor Fusion Positioning Algorithm Of GNSS/INS/ Camera Lane Line

High-precision vehicle positioning system has always been a research hotspot in academia and industry.With the development of Global Navigation Satellite System(GNSS)and Inertial Navigation System(INS)positioning technologies,it is no longer a problem for people to achieve high-precision positioning in an open environment.But in the scenarios where GNSS signals are interrupted,such as tunnels,viaducts and urban canyons,how to ensure stable,reliable,and high-precision positioning result of the system is a difficult problem in current research.Aiming at the positioning problem in the GNSS-denied scenario,it is an effective technical solution to use the visual feature information obtained from the camera sensor to combine with High-Define map(HD map)for feature matching and positioning.This paper studies the information characteristics of common vehicle sensors such as GNSS/INS/Odometer/Camera,proposes a set of camera lane line feature information to assist the GNSS / INS vehicle-mounted combined positioning system,and conducts a lot of practical tests to systematically verify the rationality and reliability of the designed multi-sensor fusion positioning algorithm.The purpose of this paper is to systematically study the theoretical model and technical methods of the lane line feature information on the auxiliary positioning of the GNSS / INS combined system,and to improve the positioning capability of the multi-sensor fusion system in the GNSS-denied scenario.The main work and contributions of the paper as follows:1)Based on the Extended Kalman Filtering(EKF)algorithm commonly used in GNSS/INS integrated navigation systems,a new lane line information assisted GNSS/INS system positioning algorithm is designed.The lane line distance information,which means the distance from the vehicle to the lane line,is used to constrain the vehicle's positioning error in the lateral direction to achieve decimeter-level positioning.We developed a GNSS/INS/Camera multi-sensor fusion positioning system,and conducted a lot of vehicle testing to quantitatively verify the effectiveness and reliability of the GNSS/INS /Camera fusion positioning algorithm.By analyzing positioning errors,it is clear that with the aid of lane line feature information,the multi-sensor fusion positioning system implemented in this paper strictly maintains the lateral positioning accuracy within 6 dm in long-term GNSS-denied scenarios.2)In order to ensure the effect of camera lane line distance information assistance and enhance the positioning accuracy of the system,the speed output information of the vehicle Control Area Network(CAN)bus and the speed constraint information of the vehicle kinematics model are analyzed and modeled without adding sensors.The derivation implements a three-dimensional speed-assisted positioning algorithm,which significantly reduces the positioning error divergence of the fusion system in the GNSS-denied scenario,and ensures the accurate matching of the lane line information recognized by the camera and the HD map lane information.In summary,the GNSS/INS/Camera multi-sensor fusion positioning algorithm proposed in this paper can effectively maintain the decimeter-level positioning accuracy,and provides a solution to overcome the problem of vehicle positioning error divergence in complex scenes such as tunnels,viaducts,and urban canyons.The scheme is one of the effective means to tap the potential value of camera information.At the same time,this solution has the advantages of low hardware cost,high precision,and strong scalability.