Myrtle becomes part of £2m collaboration into the safety and verification challenges of using deep learning in on-road vehicles

April 20, 2017

“Is it safe for a computer to drive your car?”

Cambridge software company Myrtle has become part of a £2m collaboration that will aim to accelerate the development and deployment of commercially viable, next-generation highly autonomous vehicles. Myrtle will provide algorithms and custom hardware that realizes so called “deep learning” and the collaboration will use this to identify methods for addressing the safety and verification challenges currently preventing sophisticated artificial intelligence being used in road vehicles.

Expert Collaboration

Myrtle’s CEO, Peter Baldwin, said “We’re excited and honored to be involved in a collaboration with the experts from AVLHoriba-MIRAVertizan and the manufacturing group at Warwick University”. “The current generation of self-driving car prototypes are running multiple deep learning algorithms that interact in complex ways. Not only must these algorithms be implemented in a scalable way that targets limited resources but we also need to be sure that they are safe. Such verification is a big obstacle to many commercial applications of deep learning”.

Winning Project

The winning project is called “Smart ADAS Verification and Validation Methodology (SAVVY)” and is part of a £109 million of government investment into cutting-edge automotive research and development projects overseen by Innovate UK. The project will start in July and run for two and a half  years.

Read Full Article