Making locomotive inspections verifiable

Every passenger boards a train trusting that someone has checked the locomotive carrying them. Behind every journey are people expecting to reach home safely.

Passenger locomotives undergo scheduled trip inspections according to the operator’s maintenance requirements.

Today, an inspector walks around the locomotive with a paper checklist. They examine the required parts, record their findings, and sign off. Their training and judgment are central to this work.

The checklist preserves the inspector’s conclusion, but little of what they saw while reaching it. Giving the pit manager that firsthand evidence would make findings easier to review and concerns easier to act on. It would also give the inspector a fuller record of the care they put into each check.

We are building a copilot with a helmet-mounted camera, microphone, and audio guide. It takes the inspector through one check at a time and captures their spoken findings. Vision models examine the relevant location and component, highlighting anything worth a second look. The inspector keeps the final judgment; the pit manager receives a report with the supporting evidence.

Implementation fits the wearable workflow to the operator’s checklists and inspection conditions. The harness manages the sequence of checks, evidence, and follow-up items so it can be adapted to other inspections. Inference provides the speech and vision models that help interpret what the inspector says and sees.

We are grateful to the operator for its foresight in strengthening the work its inspectors already do. One fault found in time can save lives. We hope this system helps that care reach more locomotives, more journeys, and more people on their way home.