nuReasoning is a reasoning-centric multimodal dataset for evaluating and training end-to-end driving systems. It is designed to help models connect perception, map context, actor motion, ego state, route intent, and safety constraints to interpretable driving decisions. The dataset was created by qixuewei and was last updated on May 7, 2026.
Use Cases
- Evaluate autonomous driving model performance based on multimodal driving clips.
- Train models to integrate perception and map context for driving decisions.
- Benchmark system safety and reasoning capabilities in long-tail real-world scenarios.
- Develop interpretable driving decision systems based on synchronized sensor data.
Strengths
- Focuses on reasoning and long-tail scenarios, a critical challenge for autonomous driving.
- Includes synchronized multi-camera context, LiDAR metadata, ego state, object annotations, and HD-map context.
- Designed to connect multiple driving factors (perception, motion, intent, constraints) to interpretable decisions.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
- Source
- huggingface
- Freshness
- Last updated 2026-05-07 12:05:22; freshness should be verified.