Robotics evidence guide · automated editorial preparation, not expert review
Keep human assistance visible in robot evaluations
Record interventions as part of the outcome, not as a hidden reset. A task finished with assistance is useful evidence, but it is not the same result as autonomous completion.
Evidence and interpretation
Define assistance before collecting data. It might include teleoperation, object repositioning, changing a goal or manually recovering from a stall. Maha’s proposed record asks what changed and which attempt was affected, while avoiding unnecessary operator identity. Separate the operational value of supervised assistance from a claim of autonomy. In a training dataset, demonstrations can be appropriate inputs; they must not silently become autonomous evaluation evidence. LeRobot’s episode organization gives a useful data boundary, not an answer to this classification question.
Proposed evidence workflow
- Define the intervention categories and who may declare them.
- Keep assisted attempts in the registered denominator and label their outcome.
- Report intervention counts with task context; do not infer personal worker performance.
Worked illustration
One synthetic case injects a simulated operator-assisted transfer. Both controllers record assisted completion. Removing that event changes the package and the offline verifier refuses it, even if a new package digest is calculated.
Limits
This accounting convention does not establish trust, comfort, accessibility or safety in human-robot interaction. No real operator is observed in the example.
Sources and review
Locator: What’s new in v3; Format design
The format joins time-series signals and video through episode metadata; episode boundaries are not necessarily file boundaries.
Boundary: A storage format does not establish consent, dataset coverage, annotation accuracy or model generalization.
Inspection and reuse basis
Selected sections inspected 2026-09-14. Original explanation and link; no dataset or video redistribution. This is not independent expert review of this article.