Technical papers from the work we do.
Each paper comes out of a programme we are working on, uses that programme’s real data, and records what went wrong along the way as well as what worked. They are written by the engineers who did the work.
Latest papers.
Machine Learning in the Whole Vehicle Test Community
A practical guide to integration opportunitiesMost of the value in vehicle test data sits in how the systems behave together, which diagnostics running inside a single controller cannot see. The paper makes the case for starting with anomaly detection, because it needs no labelled failures, and shows it running on ZELLOR test data.
Project ZELLOR · 23 pages · PDFAdapting New Logger Data to an Established Ingestion Workflow
A practical guide to trustworthy vehicle telemetryTwo loggers can write equally valid MDF4 files, and a pipeline built around one will still reject the other. The paper shows how we brought around 7,000 raw files from a CANedge2 logger into a workflow built for another vendor’s files, without changing that workflow, and finished with zero ingestion failures.
Project ZELLOR · 14 pages · PDFWorking on a problem like these?
Tell us where your test data is getting stuck. We’ll tell you, plainly, whether we can help.