If any of you, readers, have ever been in Jerusalem, today the capital of the State of Israel, you will understand what the American chip manufacturer Intel and Mobileye, an Israeli company that develops technologies for the detection and processing of images for the automotive, it seems like a very complex task, if not impossible.
Both companies have started the first phase of testing their joint autonomous driving project with a fleet of 100 vehicles, with the aim of demonstrating a key aspect in this type of cars of the future, safety. In the coming months, the fleet will expand to the United States and other regions. "We will demonstrate," they point out, "that our driverless cars circulate from point A to point B faster, smoother and at a lower cost than a vehicle driven by humans. And that can operate in any geography and increasing a thousand times the safety of a car driven by people. "
Why Jerusalem?
In addition to being the headquarters of Mobileye, the main reason is that the capital of Israel is one of the cities with the most extreme traffic and traffic conditions, and it is not necessary to explain why.
The software used by the autonomous cars of Intel and Mobileye has been developed using techniques of deep learning in artificial intelligence, which is governed by a formal security system called 'Security Sensitive to Responsibility' (RSS). This solution normalizes the "common sense" principles of what it means to drive safely and applies them to a set of mathematical formulas and algorithms that a machine can understand. "If the software based on artificial intelligence proposes an action that violates one of these principles of common sense, the RSS system would reject the decision," they say in Intel.
The autonomous driving system based on the artificial intelligence of the American and Israeli companies wants to become the industry standard, although there is still a long way to go, and many problems to solve.
The autonomous cars of Intel and Mobileye have 360-degree vision thanks to eight cameras, although each vehicle has 12 cameras in total that provide a long-range immersive vision. Four cameras are used exclusively for parking. The envelope vision detection system is capable of detecting other cars and pedestrians on the road, the passable routes and the semantic meaning of traffic signals and lights, and creates high-definition maps in real time. And it has the ability to locate an autonomous vehicle with an accuracy or percentage of error of only one centimeter.
Amnon Shashua, senior vice president of Intel Corporation, as well as CEO and CTO of Mobileye, notes that "Mobileye and Intel's approach to this project is contrary to the industry's common practice, which is committed to an excess in computing needs. in the R & D phase of autonomous driving, to then try to optimize, reduce costs and energy ". "We," says Shashua, "are executing a more effective strategy by subscribing fewer IT needs to maintain a greater focus on the development of more efficient algorithms for vehicle detection, driving and control."