用双向运动变换器生成逼真交通碰撞场景，提升自动驾驶测试效果
Adv-BMT: Bidirectional Motion Transformer for Safety-Critical Traffic Scenario Generation
- 通过逆时序重建交通流，从末尾状态还原初始交互
- 生成场景使自动驾驶训练后碰撞率降低20%
- 无需碰撞数据预训练，适合自动驾驶安全验证
基于场景的测试对自动驾驶系统性能验证至关重要，但现有真实世界数据集缺乏长尾安全关键场景。为解决数据稀缺问题，我们提出Adv-BMT框架，通过多样化且真实的对抗性交通交互增强真实场景。该框架核心是双向运动变换器（BMT），可执行逆向交通运动预测：以场景最后时刻的智能体信息为输入，逆时间顺序重建至初始时刻。Adv-BMT采用两阶段流程：先进行对抗性初始化，再执行逆向运动预测。不同于以往方法，本工作无需碰撞数据预训练，即可生成真实且多样的碰撞交互。实验验证了生成场景的质量：在增广数据上训练可使碰撞率降低20%。演示与代码见：https://metadriverse.github.io/adv-bmt/
原文摘要 · Abstract (English)
Scenario-based testing is essential for validating the performance of autonomous driving (AD) systems. However, such testing is limited by the scarcity of long-tailed, safety-critical scenarios in existing datasets collected in the real world. To tackle the data issue, we propose the Adv-BMT framework, which augments real-world scenarios with diverse and realistic adversarial traffic interactions. The core component of Adv-BMT is a bidirectional motion transformer (BMT) model to perform inverse traffic motion predictions, which takes agent information in the last time step of the scenario as input, and reconstructs the traffic in the inverse of chronological order until the initial time step. The Adv-BMT framework is a two-staged pipeline: it first conducts adversarial initializations and then inverse motion predictions. Different from previous work, we do not need any collision data for pretraining, and are able to generate realistic and diverse collision interactions. Our experimental results validate the quality of generated collision scenarios by Adv-BMT: training in our augmented dataset would reduce episode collision rates by 20%. Demo and code are available at: https://metadriverse.github.io/adv-bmt/.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。