LoT‑nuScenes: A Virtual Long‑Tail Scenario Dataset for Parallel Vision and Parallel Vehicles

Dataset Overview

LoT‑nuScenes is a virtual long‑tail scenario dataset for parallel vision and parallel vehicle research. The dataset constructs traffic accident scenes under different states in the CARLA simulator, including six categories of motor‑vehicle accidents and one pedestrian accident, combined with three extreme weather conditions, three time periods, and five location categories. It adopts the nuScenes format, equipped with multi‑sensor and 360° views, filling the gap in accident scenario data and achieving a long‑tail normalized distribution.

Accident Types and Annotations

The dataset includes seven accident types: rear‑end collision, lane change, wrong‑way driving, intersection collision, and sudden pedestrian crossing. Rear‑end collisions are divided into two‑vehicle rear‑end and multi‑vehicle rear‑end; lane changes are divided into collisions with neighboring vehicles and collisions with guardrails. Accident events are collected in the nuScenes format; the following shows annotation boxes for two‑vehicle rear‑end and wrong‑way driving:

Environment Requirements

• Operating System: Ubuntu 22.04
• CARLA Simulator Version: 0.9.15
• Installation Command: pip install carla=0.9.15