Intelligent driving-oriented multi-sensor fusion high-precision positioning algorithm and its optimization

With the development of intelligent driving technology, high-precision positioning has become one of the key technologies of autonomous driving. However, a single sensor has the problem of limited positioning accuracy in complex traffic environment. Therefore, this paper proposes a multi-sensor fusion positioning algorithm framework that integrates global navigation satellite system (GNSS), inertial measurement unit (IMU), LiDAR and visual sensors. The framework uses adaptive Kalman filter to estimate the state, and combines map matching optimization and robustness enhancement strategy to further improve the positioning accuracy and system stability. In addition, this paper also realizes IMU zero-bias online calibration, LiDAR motion distortion compensation, sliding window factor graph optimization and CUDA acceleration to improve the accuracy, real-time performance and robustness of the positioning system. Experiments are carried out in urban canyons, high-speed ramps and tunnels. The results show that the lateral positioning error of this algorithm is less than 20cm in complex scenes, which is significantly better than other classical algorithms, and it shows good trajectory consistency and stability in dynamic obstacle scenes, which meets the high-precision positioning requirements of autonomous driving.