What it does
All incidents and reference labels are synthetic, so agreement in this demo is not production accuracy on real robot failures.
Benchmarks & research
Triage simulated warehouse-robot incidents with Jev and compare its decisions with a small local classifier.
Screenshot unavailable. Open the experiment ↗
All incidents and reference labels are synthetic, so agreement in this demo is not production accuracy on real robot failures.
Start by listing common hardware alerts, like sensor glitches or battery warnings. Decide which repair team handles each issue and whether a person must step in immediately. Setting these clear categories and urgency levels gives your developer exact rules to follow.
Your developer can adapt the project script to send incident descriptions to Jev using a digital access key from the service. This experiment only tested pretend warehouse incidents, so you must carefully check real equipment alerts before relying on its choices.