Research On Vehicle Multi-sensor Fusion Positioning Algorithm Based On DSPACE

In order to solve the increasingly serious problems of traffic safety and congestion,autonomous driving technology has received widespread attention.Vehicle positioning is a very basic and critical part of achieving L3 and above autonomous driving.L3 and above autonomous driving vehicles need to meet the positioning accuracy of 10-30 cm.At present,the centimeter-level positioning technology of intelligent vehicle outdoors mainly depends on the RTK(Real-Time Kinematics)technology based on satellite navigation,but it cannot accept GPS signals in indoor scenes such as underground parking Lots.Therefore,vision and lidar technology are required.At this stage the vision and lidar system is immature and expensive,so there is a need to study low-cost indoor high-precision positioning technology.Ultra-wideband(UWB)positioning technology can theoretically achieve a positioning accuracy of 10 cm,which is highly accurate but susceptible to signal non-line-of-sight propagation.The strapdown inertial navigation system based on the inertial measurement unit(IMU)is a passive positioning system,which is not easily affected by the external environment,but there is error accumulation problem.This paper proposes a UWB / IMU fusion positioning method,which can provide high-precision position information for automatic navigation in indoor parking lots.This article does the following work:(1)The principle of UWB positioning system is studied.Based on the two-way time-of-flight ranging method and least squares estimation,the trilateral positioning algorithm is implemented,and the causes of UWB positioning errors are analyzed.(2)Based on the composition and principle of the inertial navigation system,the inertial navigation system position and attitude calculation algorithm is implemented and the inertial navigation system error model is established.(3)Proposed a positioning method that combines ultra-wideband and inertial navigation,constructed a mathematical model of UWB and IMU sensors,and generated simulated sensor data.Based on MATLAB,successively designed Error State Extended Kalman Filter(ES-EKF)and Unscented Kalman filter(UKF)algorithm.Simulation results verified that the proposed fusion algorithm can effectively improve the positioning accuracy and robustness.(4)An experimental platform based on dSPACE rapid control prototyping system was built,ES-EKF fusion algorithm was established using Simulink,and experimental verification was carried out on MicroAutoBox II controller.The experimental results show that the fusion positioning algorithm designed in this paper has a good effect,and its positioning accuracy is higher than that of using the ultra-wideband positioning module alone.When the base station signal is blocked,the average positioning error can be controlled at about 22 cm,which effectively improves the ultra-wideband positioning robustness of the system.This set of positioning systems can be applied to autonomous navigation and positioning of vehicles in large indoor parking lots after further integration.