Skip to main content
QUICK REVIEW

[论文解读] GRIP++: Enhanced Graph-based Interaction-aware Trajectory Prediction for Autonomous Driving

Xin Li, Xiaowen Ying|arXiv (Cornell University)|Jul 17, 2019
Autonomous Vehicle Technology and Safety参考文献 40被引用 87
一句话总结

GRIP++ 在 GRIP 的基础上通过结合固定图和可训练图来捕捉城市场景中代理之间的互动,从而在城市驾驶的轨迹预测中实现更好的性能,在 ApolloScape 上达到 state-of-the-art,且运行速度比 CS-LSTM 更快。

ABSTRACT

Despite the advancement in the technology of autonomous driving cars, the safety of a self-driving car is still a challenging problem that has not been well studied. Motion prediction is one of the core functions of an autonomous driving car. Previously, we propose a novel scheme called GRIP which is designed to predict trajectories for traffic agents around an autonomous car efficiently. GRIP uses a graph to represent the interactions of close objects, applies several graph convolutional blocks to extract features, and subsequently uses an encoder-decoder long short-term memory (LSTM) model to make predictions. Even though our experimental results show that GRIP improves the prediction accuracy of the state-of-the-art solution by 30%, GRIP still has some limitations. GRIP uses a fixed graph to describe the relationships between different traffic agents and hence may suffer some performance degradations when it is being used in urban traffic scenarios. Hence, in this paper, we describe an improved scheme called GRIP++ where we use both fixed and dynamic graphs for trajectory predictions of different types of traffic agents. Such an improvement can help autonomous driving cars avoid many traffic accidents. Our evaluations using a recently released urban traffic dataset, namely ApolloScape showed that GRIP++ achieves better prediction accuracy than state-of-the-art schemes. GRIP++ ranked #1 on the leaderboard of the ApolloScape trajectory competition in October 2019. In addition, GRIP++ runs 21.7 times faster than a state-of-the-art scheme, CS-LSTM.

研究动机与目标

  • 在交互复杂的城市环境中推动更好的自主驾驶轨迹预测。
  • 提出 GRIP++,结合固定图和可训练图,以更好地建模代理之间的交互。
  • 设计一个图卷积模型,随后是带残差连接的 Seq2Seq GRU 基轨迹预测器。
  • 在高速公路(NGSIM)和城市场景(ApolloScape)数据集上评估 GRIP++,以展示精度提升和加速效果。

提出的方法

  • 将场景表示为具有固定和可训练邻接的图,以建模代理之间的交互。
  • 应用 2D 1x1 卷积从输入速度扩展特征通道。
  • 使用交替的图操作和时序卷积层来学习时空特征。
  • 采用多块 Seq2Seq 轨迹预测器,具备编码器-解码器 GRU 和残差连接,以预测未来位置。
  • 端到端使用 Adam 训练;预测速度差分并转换为绝对位置。

实验结果

研究问题

  • RQ1将固定图和可训练图结合对城市场景中的轨迹预测精度有何影响?
  • RQ2GRIP++ 能否在像 ApolloScape 这样的城市轨迹数据集上 超越最先进方法?
  • RQ3使用速度输入和残差连接是否能改善长时域预测?
  • RQ4邻域距离阈值对预测性能有何影响?
  • RQ5在保持精度的同时,GRIP++ 与 CS-LSTM 在速度方面的比较如何?

主要发现

预测时长(s)CVV-LSTMC-VGMM + VIMGAIL-GRUCS-LSTM(M)CS-LSTMGRIPGRIP++ (Δ CS-LSTM)
10.730.680.660.690.620.610.370.38
21.781.651.561.511.291.270.860.89
33.132.912.752.552.132.091.451.45
44.784.464.243.653.203.102.212.14
56.686.275.994.714.524.373.162.94
  • GRIP++ 在 ApolloScape 轨迹数据上取得高于最先进方法的精度,超过 TrafficPredict 和 StarNet,同时改进 ADE/FDE 指标。
  • 在 NGSIM 数据集上,GRIP++ 提供具有竞争力的结果,并显示出强的短期和长期预测性能。
  • GRIP++ 在城市轨迹预测实验中比 CS-LSTM 快 21.7 倍。
  • 在消融实验中,使用固定邻域阈值 (D_close = 25 feet) 并包含附近对象可提高预测准确性,而将邻域扩展超出这一阈值可能降低性能。
  • 该模型对所有观测对象同时预测轨迹,相较于逐对象预测方案,具有效率优势。

更好的研究,从现在开始

从论文设计到论文写作,大幅缩短您的研究时间。

无需绑定信用卡

本解读由 AI 生成,并经人工编辑审核。