nuReasoning:
A Reasoning-Centric Dataset and Benchmark
for Long-Tail Autonomous Driving

Zhiyu Huang1,* Johnson Liu1,* Rui Song1,* Zewei Zhou1,2 Ruining Yang2
Yun Zhang1 Tianhui Cai1 Hanyin Zhang1 Mingxuan Gao1 Valeria Xu1
Jiali Chen1 Yishan Shen2 Yiluan Guo2 Tony (Xuewei) Qi2,† Jiaqi Ma1
*Equal contribution. Project lead. Corresponding author.
1 UCLA 2 Motional

nuReasoning contains 20K 20-second long-tail driving data clips from real-world scenarios.

Overview

nuReasoning is a large-scale real-world long-tail driving dataset containing 20K 20-second clips across diverse scenario types. The dataset provides high-quality reasoning annotations spanning spatial reasoning, driving decisions, and counterfactual reasoning. Compared with prior datasets, nuReasoning offers substantially larger-scale long-tail driving data and richer reasoning annotations, enabling models trained on it to achieve significantly improved reasoning and planning performance.

Data Schema

nuReasoning data schema
  • Clip-based dataset: 20-second self-contained real-world driving clips.
  • Rich synchronized observations: multi-view camera, Lidar point cloud, ego state, HD map, and traffic signal context.
  • Reasoning annotations: spatial reasoning, decision reasoning, and counterfactual reasoning.

Challenge

Challenge overview

Reasoning Annotations

Benchmark

Reasoning VQA Benchmark

Reasoning VQA benchmark overview

Planning Benchmark

Planning benchmark overview

Results

Reasoning VQA Results

Models trained on nuReasoning significantly improve reasoning performance across all four core capabilities in driving tasks compared to both base models and other general-purpose models.

Reasoning VQA benchmark results

Planning Results

The nuVLA model outperforms state-of-the-art AV planning methods, while training with all reasoning types (spatial, driving, and counterfactual) yields the best overall planning performance.

Planning benchmark results

Qualitative Results

Qualitative result example 1 Qualitative result example 2

BibTeX

@misc{nureasoning2026,
  title        = {nuReasoning: A Reasoning-Centric Dataset and Benchmark for Long-Tail Autonomous Driving},
  author       = {Huang, Zhiyu and Liu, Johnson and Song, Rui and Zhou, Zewei and Yang, Ruining and Zhang, Yun and Cai, Tianhui and Zhang, Hanyin and Gao, Mingxuan and Xu, Valeria and Chen, Jiali and Shen, Yishan and Guo, Yiluan and Qi, Xuewei and Ma, Jiaqi},
  year         = {2026},
  note         = {Placeholder citation.}
}