Volvo, Veoneer joint venture Zenuity teams up with CERN to speed up AV decision-making

Autonomous driving software specialist Zenuity will team up with the European Organization for Nuclear Research (CERN) to develop autonomous vehicles that can make predictions and decisions faster to help improve safety, the supplier said in a release.

This is crucial to Zenuity as its long-term goal is to help automakers create AVs that have zero collisions and cause no injuries and fatalities.

The safety of AVs has been questioned following a non-fatal accident last month in Vienna, Austria, involving a driverless bus and a pedestrian, and a fatal crash in 2018 in which an Uber Technologies self-driving test vehicle killed a pedestrian in Arizona.

When asked whether those accidents were a wake-up call for Zenuity, CEO Dennis Nobelius said the supplier doesn't need to be reminded about the importance of safety.

"Safety is in our bones. That's how we operate," Novelius told Automotive News Europe, adding that whenever an AV is involved in an accident it is a reminder of how high the stakes are in this sector. "It's imperative for the entire industry to be very cautious."

Zenuity, which is a joint venture between Volvo Cars and Veoneer, knows that a major challenge for AVs is giving them the power to accurately interpret the huge quantity of data generated by the vehicle’s cameras, lidar and radar during normal driving conditions. Scientists at CERN face a similar challenge because when the center’s Large Hadron Collider smashes sub-atomic particles this also generates a huge amount of data to assess.

"CERN wants to analyze their particle collisions in real time and we want to analyze our data coming from cameras, lidars and radar in real time," Zenuity Deep Learning Engineer Christoffer Petersson told ANE.

To address this challenge CERN uses so-called Field-Programmable Gate Arrays (FPGAs), a hardware solution that executes complex decision-taking algorithms in microseconds.

FPGAs will now be used in connection with autonomous driving.

In addition, Zenuity and CERN will collaborate on so-called “deep learning,” which is a class of machine learning algorithms. In recent years such algorithms, commonly referred to as artificial intelligence (AI), have been applied to a multitude of fields with great success, even exceeding human performance on certain tasks, the supplier said.

Zenuity hopes the collaboration with CERN will reduce the runtime and memory footprint of the relevant deep learning algorithms without reducing accuracy, while simultaneously minimizing energy consumption and cost.

Nobelius has an additional goal: "My hopes and expectations are that part of this [research] actually will go into production for the lead customer that we have for unsupervised driving, which is Volvo Cars."

When asked how quickly he wants to see that happen, Nobelius said: "A couple of years. Not longer than that."