ESP: Extro-Spective Prediction

For Long-term Behavior Reasoning in Emergency Scenarios

1Inceptio Technology, 2NJUPT, 3SHTU, 4TUM

An Emergency Scenario Emergency Icon

Left camera

Front Camera

The autonomous truck was forced to brake suddenly as a sedan cut in front of it on the highway.

Front blocking scenarios Impression Icon

On ramp merge scenarios Impression Icon

Ramp out scenarios Impression Icon

Videos of examples from the ESP dataset. The ESP-Dataset with semantic environment information focuses on emergency-event-based challenging scenarios. The collected scenarios encompassed various interactions such as merges, lane changes, ramp out, cone block, and zip lane.

Abstract

Emergent-scene safety is the key milestone for fully autonomous driving, and reliable on-time prediction is essential to maintain safety in emergency scenarios. However, these emergency scenarios are long-tailed and hard to collect, which restricts the system from getting reliable predictions.

In this paper, we build a new dataset, which aims at the long-term prediction with the inconspicuous state variation in history for the emergency event, named the Extro-Spective Prediction (ESP) problem. Based on the proposed dataset, a flexible feature encoder for ESP is introduced to various prediction methods as a seamless plug-in, and its consistent performance improvement underscores its efficacy. Furthermore, a new metric named clamped temporal error (CTE) is proposed to give a more comprehensive evaluation of prediction performance, especially in time-sensitive emergency events of subseconds. Interestingly, as our ESP features can be described in human-readable language naturally, the application of integrating into ChatGPT also shows huge potential.

Video

BibTeX

@article{dingrui2024esp,
      author    = {Wang, Dingrui and Lai, Zheyuan and Li, Yuda and Wu, Yi and Ma, Yuexin and Betz, Johannes and Yang, Ruigang and Li, Wei},
      title     = {ESP: Extro-Spective Prediction for Long-term Behavior Reasoning in Emergency Scenarios},
      journal   = {ICRA},
      year      = {2024},
    }