arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.16501cs.RO
收藏

使用Transformer网络的自动驾驶轨迹预测

Trajectory Prediction for Autonomous Driving Using a Transformer Network

University of Macau(澳门大学);School of Transportation, Southeast University(东南大学交通学院)

Zhenning Li, Hao Yu

AI总结:

本文提出基于Transformer的多模态轨迹预测框架,利用语义地图提取上下文信息,并引入惩罚偏离道路预测的辅助损失,在Lyft l5kit数据集上取得最先进性能。

摘要:

预测周围智能体的轨迹仍被视为自动驾驶中最具挑战性的任务之一。本文提出了一种基于Transformer网络的多模态轨迹预测框架。每个智能体的语义地图作为卷积网络的输入,以自动提取相关的上下文信息。本研究还提出了一种新的辅助损失,用于惩罚不可行的偏离道路预测。在Lyft l5kit数据集上的实验表明,所提出的模型达到了最先进的性能,显著提高了预测结果的准确性和可行性。

英文摘要:

Predicting the trajectories of surrounding agents is still considered one of the most challenging tasks for autonomous driving. In this paper, we introduce a multi-modal trajectory prediction framework based on the transformer network. The semantic maps of each agent are used as inputs to convolutional networks to automatically derive relevant contextual information. A novel auxiliary loss that penalizes unfeasible off-road predictions is also proposed in this study. Experiments on the Lyft l5kit dataset show that the proposed model achieves state-of-the-art performance, substantially improving the accuracy and feasibility of the prediction outcomes.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑