Learning-based algorithm attracts great attention in the autonomous driving control field, especially for decision-making, to meet the challenge in long-tail extreme scenarios, where traditional methods demonstrate poor adaptability even with a significant effort. To improve the autonomous driving performance in extreme scenarios, specifically consecutive sharp turns, three deep reinforcement learning algorithms, i.e. Deep Deterministic Policy Gradient (DDPG), Twin Delayed Deep Deterministic policy gradient (TD3), and Soft Actor-Critic (SAC), based decision-making policies are proposed in this study. The role of the observation variable in agent training is discussed by comparing the driving stability, average speed, and consumed computational effort of the proposed algorithms in curves with various curvatures. In addition, a novel reward-setting method that combines the states of the environment and the vehicle is proposed to solve the sparse reward problem in the reward-guided algorithm. Simulation results from the road with consecutive sharp turns show that the DDPG, SAC, and TD3 algorithms-based vehicles take 367.2, 359.6, and 302.1 ​s to finish the task, respectively, which match the training results, and verifies the observation variable role in agent quality improvement.

Autonomous driving is a promising way to future safe, efficient, and low-carbon transportation. Real-time accurate target detection is an essential precondition for the generation of proper following decision and control signals. However, considering the complex practical scenarios, accurate recognition of occluded targets is a major challenge of target detection for autonomous driving with limited computational capability. To reveal the overlap and difference between various occluded object detection by sharing the same available sensors, this paper presents a review of detection methods for occluded objects in complex real-driving scenarios. Considering the rapid development of autonomous driving technologies, the research analyzed in this study is limited to the recent five years. The study of occluded object detection is divided into three parts, namely occluded vehicles, pedestrians and traffic signs. This paper provided a detailed summary of the target detection methods used in these three parts according to the differences in detection methods and ideas, which is followed by the comparison of advantages and disadvantages of different detection methods for the same object. Finally, the shortcomings and limitations of the existing detection methods are summarized, and the challenges and future development prospects in this field are discussed.