Edge specialist Klas has launched RAVEN, a compute and data platform for self driving and ADAS R&D.

RAVEN is designed to provide developers with all the hardware required to ingest and store massive amounts of in-vehicle data, along with the connectivity and compute capabilities they need to turn terabytes of data into high-performing AV algorithms.

RAVEN includes: GPU support for compute intensive AV applications with virtualisation support to run any AV software; storage for up to 240TB of in-vehicle data in a swappable cassette with integrated compute for processing data in real time; connectivity to CAN, Ethernet, and 3G/4G/5G with support for software-defined networking to enable secure over-the-air access from any location; and secure remote data management, provisioning, and over-the-air updates with the ability to automate management through Ansible.

“Accelerating the development of AVs requires the building and maintenance of fleets of test cars. This requires numerous pieces of hardware that all have to fit together into one system with data collection. RAVEN simplifies that entire process,” said Frank Murray, CTO of Klas. “With RAVEN, OEMs and Tier 1s get a flexible, modular solution with the key functions of datalogging, high performance compute, and connectivity all in a standard 19″ rack format.”

“Self-driving cars must be trusted to perform in all road conditions and keep up with constantly evolving regulations, which is why it’s critical to have a scalable solution that maximises the number of test vehicles on the road at any given time. Instilling confidence in both regulators and consumers hinges on having evidence-based testing data collected at scale,” said John Gallagher, CEO of Klas. “Historically, this has been difficult to do because of reliability and performance concerns along with size and power constraints, and we’ve solved both of those issues with RAVEN. It’s rugged, compact, and truly built for the open road.”