Developing self-driving cars is the automotive equivalent of putting a man on the moon, says Itay Gat, a key player at Mobileye, the Israel-based company that is developing collision detection and driver assistance systems with automakers such as BMW and General Motors. Gat expects fully autonomous vehicles to be a reality within five years. He also told Automotive News EuropeCorrespondent Olive Keogh that demand for driver assistance solutions is well ahead of projections.
Industry estimates suggest the demand for assistance components is rising by a third annually. Is that in line with your view?
Actually we see even a bigger demand. Our estimate is that in the coming few years we will see a rise of something like 46 percent per year due to the fact that regulators are saying these kinds of systems should be standard features.
Everybody wants to know when fully automated driving is coming. What is your opinion?
Fully autonomous driving, meaning the car performs everything that we feel is needed, is expected to hit the market in the 2020-21 timeframe.
You have been working closely with BMW and Intel on autonomous driving. How significant is that partnership?
It is very significant because we see the effort of bringing such a solution to the market as one that necessitates very tight collaboration with the car manufacturers. To a large extent providing full autonomous driving is a big leap compared to what we have now and is similar to sending a man to the moon.
Has your relationship with automakers changed because autonomous driving requires earlier involvement in the development process?
Yes, that's exactly right. While we have had very good and productive relationships with the car manufacturers throughout the year [the relationship] definitely went through an upgrade when talking about the collaborations necessary to provide autonomous vehicles.
Why is autonomous driving such a complex technical challenge?
With a regular technological development it is possible to make adjustments and changes along the way and to fix things in the next version. When we are talking about autonomous driving you don't have that luxury. You must have the best possible solution from the day the system is launched on the road. This requires massive testing, verification and integration over a number of years.
How seriously do you take the threat of hacking?
Very seriously. We were thinking about this issue from the inception of the Mobileye EyeQ chip and created a situation where it would not be communicating with the outside world directly. In addition, even if some element of the system is hacked there are other security measures in place so it would not affect the communication between EyeQ and these other elements in the vehicle. We see this as critical because of the influence it could have on autonomous driving.
Given the short timeframe, will there be markets where autonomous driving systems will not be activated because of deficiencies in infrastructure or legislation?
The legislation is coming along and once the system's safety is proved, and that is exactly what we are working to show now, it will be less of a hurdle. Current infrastructures would definitely be sufficient because the performance level we are aiming for with autonomous driving does not by itself necessitate further infrastructure investment. The type of fully autonomous vehicles we are building now would be self-sufficient based on today's capabilities.
How long will it be before systems such as collision avoidance will be standard on smaller vehicles?
The very good news is that such systems are available and will be a standard fit not only on luxury cars, but basically on every car very soon. These systems provide substantial value to consumers in terms of safety. And regulators in several regions are beginning to provide credit during crash testing for standard fit collision avoidance systems, to the point where it will be difficult to get an excellent safety rating without the technology.
