CERN scientists join Zenuity machine learning for autonomous vehicles initiative

Zenuity, a joint venture between Volvo and Autoliv, today said it plans to work with European Organization for Nuclear Research (aka CERN) to speed machine learning for autonomous vehicles.

Based near Geneva, Switzerland, CERN works with scientists from 23 EU member states and is best known as maker of the Hadron Particle Collider, a 16-mile underground loop that pushes subatomic particles close to the speed of light.

Zenuity said it’s working with CERN to address big-data challenges introduced by autonomous vehicles. Cars that compute quickly can reach decisions faster and potentially avoid accidents.

“Addressing these issues is crucial for the development of safe autonomous driving (AD) cars and is a key part of Zenuity’s long-term ambition to speed up the development of vehicles that will completely eliminate car collisions and associated injuries and fatalities,” a Zenuity blog post said.

Between cameras and lidar sensors that generate megabytes of data each second and GPS, radar, and other sensors that continuously generate data, Intel predicted in 2016 that autonomous vehicles will grow to generate 4,000 gigabytes of data each day by 2020.

Given the great compute resources needed to power the particle collider and its nuclear research, CERN is experienced with working with large amounts of data.

“I think it says something important about the collaborative nature of science that an organization like CERN that conducts high-energy particle collisions can work with a company that is dedicated to completely eliminating collisions … in traffic,” said Zenuity CEO Dr. Dennis Nobelius in a statement.

Along with TPUs and GPUs, FPGAs are part of the acronym alphabet soup of hardware that enables machine learning acceleration such as Microsoft’s Project Brainwave. FPGAs are also used in high-performance computing operations and datacenters. For example, Azure uses FPGAs for encryption and compression in all its datacenters, and Intel’s Agilex line of chips that focus on datacenter operations released its first chip today.

Zenuity and CERN want to use FPGAs as part of the goal to better handle big data, but also to reduce memory consumption and runtime needs of deep learning algorithms.

Based in Gothenburg, Sweden, Zenuity currently has more than 600 employees in China, Germany, Sweden, and the United States.