Friday, September 24

A Montreal brain for the Mercedes-Benz autonomous car

For the truly autonomous car to see the light of day, international standards will be needed. The Mercedes-Benz group has therefore created a consortium called AI-SEE whose objective will be to propose standards to the European Union. And it is the Montreal company Algolux that has the role of creating the “brain” of the entire operation.

Algolux and its hundred employees spread between Montreal, Palo Alto, in California, and Munich, in Germany, specialize in algorithms which allow an autonomous system to “see” the road well from the various sensors it uses. (lidars, radars, sonar, etc.), even in a situation of reduced visibility, such as in winter, for example. The twenty AI-SEE partners are giving themselves three years to produce a system capable of driving a car without any human intervention, regardless of the season and the weather. In the jargon of the automotive sector, we are talking about a level 4 autonomous car, on the five levels of advancement of this technology.

At level 5, the vehicle has no steering wheel or pedals. The most modern vehicles currently on the road, and having their own advanced driving assistance system, are not quite at level 3, where you could let go of the wheel without worrying for several tens of seconds. .

Despite promises from the many major manufacturers who saw autonomous vehicles on the road as early as 2019, the technology is slow to see the light of day. It is complex, this technology, and is not unanimous. Should we use specific sensors? Simple cameras? Will it be necessary to incorporate the data provided by the smart furniture of the large urban centers of tomorrow?

A standard to be defined

“For the autonomous car to work, it will be necessary to agree on an industry standard”, according to Allan Benchetrit, CEO of Algolux. The technology already exists, he recalls. What Mercedes-Benz wants to do with Algolux is to create a standardized level 4 system that will prevent all manufacturers from carrying out all the stages in the design of such a system. “We must accelerate its development so that we can come to its commercialization,” he said.

Standardization will also make autonomous vehicles more affordable. The other reason why we don’t see many of them on the roads is that they are currently very expensive to produce. If Waymo, the Google subsidiary which already has a few dozen Chrysler-branded vans equipped with its autonomous driving technology on the road, were to sell its vehicles to the public, they would cost at least $ 150,000, nearly three times the price of a classic van.

Very few motorists, if not absolutely none, will pay this amount for such a vehicle. This is why Waymo and other North American companies specializing in this sector, such as Cruise, a subsidiary of the General Motors group, believe that their future will take the form of a fleet of robotic taxis that will be hailed by individuals via their mobile, like the Uber car-sharing service, but without a driver on board.

Fleets of robot taxis circulating in specific neighborhoods in a few large cities around the world could therefore be the first mainstream incarnation of autonomous vehicles, believes Mr. Benchetrit. “There is a performance urgency on the side of these companies which suggests that their autonomous vehicle technology will be on the road very soon. “

By shuttle!

Robot taxis will be just one example of the technology’s application, according to Algolux. The Montreal SME is particularly well placed to compare the different applications that the industry wishes to make of the technology. It is the only North American company to participate in the AI-SEE consortium. Ironically, the Montreal company exists thanks to financing in which the General Motors group has participated twice. Its European office has been set up in Munich in the hope of getting closer to the Bavarian manufacturer BMW. And in the end, the expert in artificial intelligence and in computer vision applied to the automotive sector will be a partner of Mercedes-Benz.

The German brand belongs to a group called Daimler, which also manufactures heavy trucks. It is on this side that we will probably see a first real generalized use of driverless piloting, believes Allan Benchetrit. “The discourse about technology has evolved a lot in recent years, to focus more on the security aspect,” he says. We can see its potential in commercial vehicles and public transit which are on the road several hours a day, every day, to reduce the workload of truckers.

Shuttles that constantly travel the same route, such as buses that link hotels and train, bus or airport stations, are the target of manufacturers. These predict the arrival of level 4 shuttles on the road before the end of the decade. Maybe even as early as 2024 or 2025. In any case, this is the deadline that AI-SEE sets itself.

Watch video



Reference-feedproxy.google.com

Leave a Reply

Your email address will not be published. Required fields are marked *