智能科学与技术学报 ›› 2023, Vol. 5 ›› Issue (2): 180-199.doi: 10.11959/j.issn.2096-6652.202317

• 综述与展望 • 上一篇    下一篇

基于深度学习的自动驾驶多模态轨迹预测方法:现状及展望

黄峻1,2,3, 田永林1,4, 戴星原1,5, 王晓3,6,7,8, 平之行4,5,9   

  1. 1 中国科学院自动化研究所复杂系统管理与控制国家重点实验室,北京 100190
    2 澳门科技大学,澳门 999078
    3 青岛智能产业技术研究院,山东 青岛 266000
    4 北京怀柔平行传感智能研究院,北京 101400
    5 中国科学院自动化研究所北京市智能化技术与系统工程技术研究中心,北京 100190
    6 安徽大学人工智能学院,安徽 合肥 230601
    7 自主无人系统技术教育部工程研究中心,安徽 合肥 230601
    8 安徽省无人系统与智能技术工程研究中心,安徽 合肥 230601
    9 北方自动控制技术研究所,山西 太原 030006
  • 修回日期:2023-05-26 出版日期:2023-06-15 发布日期:2023-06-10
  • 作者简介:黄峻(1998- ),男,中国科学院自动化研究所复杂系统管理与控制国家重点实验室轨迹规划工程师,主要研究方向为智能交通系统、深度学习、强化学习
    田永林(1994- ),男,博士,中国科学院自动化研究所博士后,主要研究方向为计算机视觉、智能交通
    戴星原(1993- ),男,博士,中国科学院自动化研究所助理研究员,主要研究方向为人工智能、智能交通系统、认知自动驾驶
    王晓(1988- ),女,博士,安徽大学人工智能学院教授,青岛智能产业技术研究院院长,主要研究方向为社会交通、动态网群组织、平行智能和社交网络分析
    平之行(1961- ),男,北方自动控制技术研究所资深研究员,主要研究方向为平行系统、平行智能、社会计算、知识自动化
  • 基金资助:
    国家自然科学基金项目(62173329)

Deep learning-based multimodal trajectory prediction methods for autonomous driving: state of the art and perspectives

Jun HUANG1,2,3, Yonglin TIAN1,4, Xingyuan DAI1,5, Xiao WANG3,6,7,8, Zhixing PING4,5,9   

  1. 1 The State Key Laboratory for Management and Control of Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China
    2 Macau University of Science and Technology, Macau 999780, China
    3 Qingdao Academy of Intelligent Industries, Qingdao 266000, China
    4 Beijing Huairou Academy of Parallel Sensing, Beijing 101400, China
    5 Beijing Engineering Research Center of Intelligent Systems and Technology, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China
    6 School of Artificial Intelligence, Anhui University, Hefei 230601, China
    7 Engineering Research Center of Autonomous Unmanned System Technology, Ministry of Education, Hefei 230601, China
    8 Anhui Provincial Engineering Research Center for Unmanned System and Intelligent Technology, Hefei 230601, China
    9 North Automatic Control Technology Institute, Taiyuan 030006, China
  • Revised:2023-05-26 Online:2023-06-15 Published:2023-06-10
  • Supported by:
    The National Natural Science Foundation of China(62173329)

摘要:

对周围车辆轨迹的精确预测可以辅助自动驾驶车辆做出合理的即时决策。虽然相比传统轨迹预测算法,深度学习方法已取得较好效果,但是自动驾驶车辆在异构高动态复杂变化环境下实现多模态高精度预测仍存在信息丢失、交互和不确定性难以建模、预测缺乏可解释性等问题。Transformer具备的长距离建模能力和并行计算能力使其不仅在自然语言处理领域取得巨大成功,而且在扩展至自动驾驶多模态轨迹预测任务时也解决了以上问题。基于此,对过去基于深度神经网络的方法,特别是对基于Transformer的方法进行全面总结与回顾;同时分析了Transformer相较于传统序列网络、图神经网络、生成模型的优势,并结合现有难题进行针对性分析与分类。Transformer模型可以更好地应用于多模态轨迹预测任务,此类模型具有更好的泛化性和可解释性。最后,对多模态轨迹预测未来发展方向进行了展望。

关键词: Transformer, 序列网络, 图神经网络, 生成模型, 轨迹预测, 多模态

Abstract:

Although deep learning methods have achieved better results than traditional trajectory prediction algorithms, there are still problems such as information loss, interaction and uncertainty difficulties in modelling, and lack of interpretability of predictions when implementing multimodal high-precision prediction for autonomous vehicles in heterogeneous, highly dynamic and complex changing environments.The newly developed Transformer's long-range modelling capability and parallel computing ability make it a great success not only in the field of natural language processing, but also in solving the above problems when extended to the task of multimodal trajectory prediction for autonomous driving.Based on this, the aim of this paper is to provide a comprehensive summary and review of past deep neural network-based approaches, in particular the Transformer-based approach.The advantages of Transformer over traditional sequential network, graphical neural network and generative model were also analyzed and classified in relation to existing challenges, simultaneously.Transformer models can be better applied to multimodal trajectory prediction tasks, and that such models have better generalisation and interpretability.Finally, the future directions of multimodal trajectory prediction were presented.

Key words: Transformer, sequential network, graph neural network, generative model, trajectory prediction, multimodal

中图分类号: 

No Suggested Reading articles found!