汽车工程 ›› 2024, Vol. 46 ›› Issue (7): 1219-1227.doi: 10.19562/j.chinasae.qcgc.2024.07.009

• • 上一篇    

引入自监督预训练的轨迹预测方法

李琳辉,付一帆,王霆,王雪成,连静()   

  1. 大连理工大学机械工程学院,大连 116024
  • 收稿日期:2024-01-23 修回日期:2024-03-02 出版日期:2024-07-25 发布日期:2024-07-22
  • 通讯作者: 连静 E-mail:lianjing@dlut.edu.cn
  • 基金资助:
    国家自然科学基金(52172382);中央高校基本科研业务费项目(DUT22JC09);辽宁省科学技术计划项目(2022JH1/10400030)

Trajectory Prediction Method Enhanced by Self-supervised Pretraining

Linhui Li,Yifan Fu,Ting Wang,Xuecheng Wang,Jing Lian()   

  1. School of Mechanical Engineering,Dalian University of Technology,Dalian 116024
  • Received:2024-01-23 Revised:2024-03-02 Online:2024-07-25 Published:2024-07-22
  • Contact: Jing Lian E-mail:lianjing@dlut.edu.cn

摘要:

针对目前基于监督学习的轨迹预测模型数据利用效率低、精度有限的问题,提出一种轨迹预测模型及通用的自监督预训练策略。首先,基于Transformer搭建轻量化的轨迹预测模型,实现场景时序空间特征提取与交互关系建模;其次,设计运动信息时序掩码、道路信息空间掩码、交互关系掩码3类掩码重建任务对模型进行自监督预训练,以提升模型对场景通用特征的提取能力;最后,以预训练权重为初始化参数在下游任务中进行监督学习微调。在Argoverse2 Motion Forecasting数据集的实验表明,模型在预训练任务中能够很好地重建出交通场景,引入自监督预训练能够有效提升预测精度和数据利用效率,且对不同预测任务具有通用性,在单目标轨迹预测与多目标轨迹预测任务上minFDE6指标分别提升3.3%与3.7%。

关键词: 自动驾驶, 轨迹预测, 自监督预训练

Abstract:

To address limitation in prediction accuracy and data utilization efficiency of supervised learning-based trajectory prediction models, a trajectory prediction model and a general self-supervised pretraining strategy are proposed. Firstly, a lightweight trajectory prediction model based on Transformer is established to extract temporal-spatial features while modeling interaction relationship. Secondly, three types of masks, namely motion information temporal mask, road information spatial mask, and interaction relationship mask, are designed for self-supervised pre-training tasks on the model to enhance the model's ability to extract general scene features. Finally, pretraining weights are used as initialization parameters for supervised learning fine-tuning in downstream tasks. Experimental results on the Argoverse2 Motion Forecasting dataset show that the model can effectively reconstruct traffic scenes in pretraining tasks. The introduction of self-supervised pretraining improves prediction accuracy and data utilization efficiency. Moreover, it exhibits universality for different prediction tasks, achieving a 3.3% and 3.7% improvement in the minFDE6 for single-agent and multi-agent trajectory prediction tasks, respectively.

Key words: autonomous driving, trajectory prediction, self-supervised pretraining