Autonomous Driving | Recent Advancements in End-to-End Autonomous Driving using Deep Learning: A Survey

• code：https://github.com/Pranav-chib/End-to-End-Autonomous-Driving

1. Introduction

• 端到端自动驾驶方向研究的演化趋势，逐年递增；

  

  端到端自动驾驶方向研究的演化趋势

• 首先，研究专注于解决误差传播（error propagation），单个任务的学习；

  • Neat: Neural attention fields for end-to-end autonomous driving, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 15793– 15803.
  • Trajectory-guided control prediction for end-to-end autonomous driving: A simple yet strong baseline, in: NeurIPS, 2022.

• 随后，端到端模型通过设置task-specific outputs，体现了计算上的优势

• 端到端模型提升可解释性的一些手段：

  • auxiliary outputs
    • Transfuser: Imitation with transformer-based sensor fusion for autonomous driving, IEEE Transactions on Pattern Analysis and Machine Intelligence (2022)
    • Trajectory-guided control prediction for end-to-end autonomous driving: A simple yet strong baseline, in: NeurIPS, 2022.
  • attention maps
    • Multi-modal fusion transformer for end-to-end autonomous driving, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 7077–7087
    • Policy pre-training for end-to-end autonomous driving via self-supervised geometric modeling, arXiv preprint arXiv:2301.01006 (2023).
    • Reasonnet: End-to-end driving with temporal and global reasoning, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 13723–13733.
    • Scaling selfsupervised end-to-end driving with multi-view attention learning, arXiv preprint arXiv:2302.03198 (2023).
    • Plant: Explainable planning transformers via object-level representations, in: CoRL 2022 Workshop on Learning, Perception, and Abstraction for Long-Horizon Planning.
  • interpretable maps [18, 19, 8, 20, 12, 21]
    • Think twice before driving: Towards scalable decoders for end-to-end autonomous driving, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 21983–21994
    • St-p3: End-to-end vision-based autonomous driving via spatial-temporal feature learning, in: Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part XXXVIII, Springer, 2022, pp. 533–549
    • Human-ai shared control via policy dissection, Advances in Neural Information Processing Systems 35 (2022) 8853–8867.
    • Neat: Neural attention fields for end-to-end autonomous driving, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 15793– 15803

• 端到端增加了防御敌对攻击的可能（例如操纵传感器输入等）

• 一些相关的自动驾驶技术调研文章：

  • A survey of autonomous driving: Common practices and emerging technologies, IEEE access 8 (2020) 58443–58469.
  • A survey on imitation learning techniques for end-to-end autonomous vehicles, IEEE Transactions on Intelligent Transportation Systems 23 (9) (2022) 14128–14147.
  • A survey of deep rl and il for autonomous driving policy learning, IEEE Transactions on Intelligent Transportation Systems 23 (9) (2021) 14043–14065
  • A survey of end-to-end driving: Architectures and training methods, IEEE Transactions on Neural Networks and Learning Systems 33 (4) (2020) 1364–1384.
  • End-to-end autonomous driving: Challenges and frontiers, arXiv 2306.16927 (2023).

• 

  文章的内容结构情况

• 

  感知部分的处理流程

总结

文章讲得太泛了，等有空了再看