Intel and Mobileye talk up new fleet for Level 4 autonomous driving

To celebrate the completion of the completion of Intel’s tender offer for outstanding ordinary shares of Mobileye, the pair have announced a new initiative to build a fleet of 100 autonomous vehicles.

The fleet itself will be Level Four on the SAE International scale of autonomous driving, which is almost as good as it gets. The first cars will hit the road by the end of the year, testing throughout the US, Europe and Israel, using multiple car brands and vehicle types.

When it comes to the development of autonomous vehicles, Level Four is pretty advanced (there are only five levels). It essentially means that the vehicle can operate without the intervention of a human in almost every scenario. Perhaps the only environment it will struggle with is a busy city centre, but autonomous driving in this sense is still decades away and probably requires mass market adoption of the vehicles.

“Building cars and testing them in real-world conditions provides immediate feedback and will accelerate delivery of technologies and solutions for highly and fully autonomous vehicles,” said Amnon Shashua, who will lead the newly integrated autonomous vehicles business at Intel.

“Geographic diversity is very important as different regions have very diverse driving styles as well as different road conditions and signage. Our goal is to develop autonomous vehicle technology that can be deployed anywhere, which means we need to test and train the vehicles in varying locations.”

He’s not wrong either. Technology is crucial in the development of autonomous vehicles, but perhaps experience is just as important. If you have the best technology in the world, but no means to collect data then it is ultimately useless.

While it might seem obvious, it is worth highlighting every now and then, as the obvious things tend to get forgotten; the only way autonomous vehicles are going to be a success is through millions of hours of road testing. This is the way in which the AI component of the car will learn; machine learning at its most basic level is the idea of experiencing events, comparing the outcome to similar events, and deciding which was better for similar scenarios in the future. Due to the number of variables on the roads, the more experience the better.

The autonomous car landscape is already starting to get pretty congested, so this could be a clever idea from Intel. As well as demonstrating the hybrid-solution, the team will also be collected valuable information.

“This does not replace any customer activities; it is additive to them,” said Shashua. “Our customers will benefit from our ability to use this fleet to accelerate our technology development. We want to enable automakers to deliver driverless cars faster while reducing costs – data we collect will save our customers significant costs.”

For those customers who choose Intel as a partner, they don’t just get the technology, but with this initiative, there is some independent data as well. Without prompt, Intel is collecting important information which can help train an AI application to be a better driver. Depending on how many miles the new Intel fleet gets under its tyres, this could be a nice little value add in the raging race for dominance on the roads.