2018 VW - Mobileye
The picture shows Professor Amnon Shashua, Technical Director and Chairman of Mobileye and Dr. Herbert Diess, CEO of VW, at the signing of a strategic partnership declaration at the Consumer E
lectronics Show in Las Vegas in January 2018.
Mobileye is an Israeli company, owned by Intel since March 2017 for $ 15.3 billion. Wikipedia counts over 100,000 employees for Intel and alone in the PC microprocessors, a market share of 80 percent. Mobileye N.V.
exists since 1999 and is a world leader in the development of computer vision and machine learning, data analysis, localization and mapping for advanced assistance systems and autonomous driving.
The software associated with the cameras can already be found today in more than 24 million vehicles of various manufacturers, including 9 million in 2017. Such sensor technology forms the basis for driver
assistance, e.g. emergency braking assistance, lane tracking, traffic sign recognition and much more.
In 2017, they began to build a series of 100 semi-autonomously driving cars at least reaching level 4. Hardly to be recognized from the outside the 12 cameras, radar sensors and laser scanner as redundancy. The
cameras have a high resolution and range from 250 to 300 meters, radar about twice so wide. Sensing results in much more accurate maps than previous ones, being then stored in a cloud as so-called swarm data.
Mobileye will be able to exert a tremendous influence on how widespread and well-organized these data are and how they are organized. Problems are less the data acquisition, but their interpretation. After all, a
detailed model of the environment is to be saved afterwards.
How does the software recognize even with high-resolution and 360° of view at sufficient brightness a woman with a stroller in the blind spot of a truck that threatens to be run over when it turns? It would be the basic
requirement for everything else, e.g. to warn the driver. So this should happen automatically, because you can not always trust that all image information is considered.
A typical requirement for assisted driving. Although the number of accidents has dropped dramatically in the last 50 years, more people still die outside than inside a vehicle. The system must therefore as quickly as
possible assign image data to at least a two-digit number of elements, a typical requirement of artificial intelligence. Supposedly, this is already possible to almost 95 percent.
It is important to mention that despite the fact that the daylight and the weather are very decisive here, one hundred percent safety can not be achieved by evaluating the sensors of cameras. The first, also by Mobileye
pursued approach is the information of additional sensors (redundancy). The example shows how important here, too, could be the swarm data from other road users and at risk points of permanently installed
sensors.
Meanwhile, it is controversial, potentially in traffic endangered as the woman with stroller itself electronically to attract attention. The data-legal side and the possible failure of such a signal that others may rely on. Good
chances for worldwide almost unique speed in the evaluation and the data transmission seems to have Mobileye as part of Intel anyway. 06/18
| 
