Research On Positioning Methods Based On GNSS/IMU/LIDAR Multi-source Information Fusion

Navigation and positioning systems to achieve stable and accurate positioning in complex environments is a prerequisite for intelligent behaviour and autonomous decision-making on unmanned mobile platforms.Currently,a single system such as a global satellite navigation system or inertial navigation system configured on an unmanned mobile platform can no longer meet the needs of multi-scene navigation.To solve this problem,multiple types of sensors are introduced and the redundant information they acquire is fused to build a more stable and effective multi-sensor information fusion positioning system is a hot issue in current research.In this paper,we select consumer-grade GNSS,IMU and Li DAR sensors,analyse the mathematical models and error characteristics of the three types of sensors,study multi-source information fusion algorithms,build a combined navigation/Li DAR odometry fusion positioning system based on graph optimization,and design experiments to verify the feasibility and stability of the system,the main work is as follows:(1)According to the GNSS positioning principle and IMU's navigation projection model,the respective advantages,disadvantages and error characteristics of the two systems are analysed,the performance characteristics of consumer-grade GNSS and IMU are combined,the method of determining the sensor noise term is studied,and the GNSS/IMU combined navigation system is implemented by applying the extended Kalman filter.The combined navigation experiments were also conducted.In the unobstructed environment,the combined GNSS/IMU navigation system has high positioning accuracy;in the GNSS signal deficient environment,the combined navigation system generates cumulative errors and its positioning accuracy decreases.(2)The ICP and NDT point cloud matching algorithms are investigated based on the characteristic point cloud data of Li DAR.In this paper,a LIDAR odometer is constructed for consumer-grade LIDAR with ICP as the point cloud matching algorithm.It is proved through experiments that the LIDAR odometer has high navigation and positioning accuracy in a short period of time,but long-time operation will cause the accumulation of matching errors and positioning deviations.(3)To address the above-mentioned problems of combined navigation system and LIDAR odometer,this paper proposes a method for fusing GNSS/IMU combined navigation and LIDAR odometer for positioning based on graph optimization theory.The method overcomes the limitations of filter-based fusion methods in non-linear systems,and uses the LIDAR odometer state information as the quantity to be optimised,and optimises the odometer state quantity by combining the navigation output values and the observation equation between the two,to construct a graph optimisation model.(4)By building a hardware and software experimental platform,the absolute trajectory error,root mean square error and other indicators of multiple fused positioning modes(IMU-LIDAR,GNSS-LIDAR,GNSS-IMU-LIDAR and EKF-LIDAR)were analysed.The experimental results show that the EKF-LIDAR with combined navigation and LIDAR odometry fusion has the best positioning performance,and the two The joint optimization function is constructed to correct for the accumulated error and find the optimal estimate.In the experiments where GNSS is interrupted in the simulated occlusion environment,the normal operation of the system can still be guaranteed for a short time.When GNSS information is restored,the cumulative error of the system is quickly updated to normal values.