New Integrated Multi-Algorithm Fusion Localization and Trajectory Tracking Framework of Autonomous Vehicles under Extreme Conditions with Non-Gaussian Noises

This paper proposes a novel integrated multi-algorithm fusion localization and trajectory tracking framework for autonomous vehicles under extreme conditions with non-Gaussian noises. Firstly, in order to solve the problem that GPS signals are interfered with non-Gaussian noises or lost, a localization method based on Particle Filter (PF) is designed, which takes full advantage of the reference objects position information and vehicle driving state information, thus realizing the self-localization for high-speed autonomous vehicles. Besides, considering the accumulated errors of the model-driven Inertial Measurement Unit (IMU) in the long-horizon positioning prediction, an online future driving state prediction algorithm based on multi-order variable-step Markov model (MM) is proposed to calculate the future vehicle position in scenarios without reference. The fusion of these two methods can give full play to their respective advantages, thus improving the accuracy and robustness of the whole localization algorithm in scenes with non-Gaussian noises. Then, the location information and the future driving state are applied to the trajectory tracking controller based on adaptive model predictive control (AMPC). Finally, the CarSim-Matlab/Simulink cGAOo-simulations results show the effectiveness of the proposed framework when GPS signal is interfered with non-Gaussian noises, which further improve the positioning accuracy and autonomous tracking stability.