BATON-Sample: Bidirectional Human-Vehicle Control Transitions in Naturalistic Driving

A sample release of the BATON benchmark, submitted to ACM Multimedia 2026, provides data for exploring human–DAS control transitions. It includes 43 routes covering 9 modalities, intended for quick exploration and prototyping. The full dataset contains 380 routes and is authored by HenryYHW.

Use Cases

• Modeling driver behavior during control transitions based on multimodal sensor data.
• Benchmarking autonomous driving systems on human-in-the-loop scenarios.
• Analyzing driver operational patterns in naturalistic settings based on behavioral data.
• Prototyping multimodal fusion algorithms for vehicle cabin and external environment data.

Strengths

• Sample includes 43 routes for initial exploration.
• Covers 9 different data modalities.
• Full dataset scale is 380 routes.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Collection Method

Likely collected from naturalistic driving studies.

Time Range

Submitted for publication in 2026.

Freshness

Last updated 2026-04-09 03:32:32; freshness should be verified.

Geography

Spatial coverage is not specified.