Reproducible guides

Validate robot data without overstating the evidence.

Each guide names the tool version, command, artifact, coverage, and unresolved decisions. These are runnable procedures—not customer case studies or claims about an uninspected dataset.

Validate a LeRobot Dataset Before ACT or SmolVLA Training

Run a local, full-integrity LeRobot readiness check and preserve a machine-readable artifact before starting ACT or SmolVLA training.

openbot-data 0.0.3 · act.readiness.json (`openbot.dataset_readiness.v1`)

Open guide

Why a Sample Audit Is Not Full Dataset Readiness

Compare sampled and full-integrity robot dataset audits without overstating what a fast preflight actually verified.

openbot-data 0.0.3 · sample.audit.json and full.audit.json (`openbot.dataset_audit.v1`)

Open guide

Robot Dataset License and Provenance Checklist

Record source identity, access terms, format evidence, and unresolved license questions before adopting a robot dataset.

openbot-data 0.0.3 · source.audit.json (`openbot.dataset_audit.v1`)

Open guide

LeRobot vs RLDS vs MCAP for Robot Data

Choose a robot-data representation by workflow stage instead of treating LeRobot, RLDS, and MCAP as interchangeable file formats.

Decision guide · 3 reviewed sources

Open guide

When Egocentric Data Helps Robot Learning

Separate the useful perception and task signals in first-person human video from the robot-native supervision it usually does not contain.

Decision guide · 3 reviewed sources

Open guide

How to Select a Robot Dataset for VLA Training

Choose VLA data by target behavior, embodiment, action semantics, observation signals, rights, and evaluation evidence—not record count alone.

Decision guide · 3 reviewed sources

Open guide