The future is near and always a big topic. Whether mobility, health or safety: people should be relieved, live healthier and particularly safe. Companies are using ever newer technologies that are becoming more sophisticated and should learn as much as possible. So does Continental. The German automotive supplier has set itself the task of making the roads safer. Not only with increasingly modern tires, but now also with an intelligent vehicle system that is supposed to make automated driving even more precise and therefore safer. The basic requirement for such highly developed driver assistance functions and automated driving is a detailed understanding of the complete traffic situation and its precise assessment.
Greetings from Matrix
In order for the vehicle to take over the driver's tasks, this must develop an understanding of all upcoming actions by all road users in order to always be able to make the right decision in every individual traffic situation. Deep machine learning methods are to be used to train algorithms that are supposed to master this task. At CES Asia, Continental is now showing a computer vision platform that uses artificial intelligence (AI), neural networks and machine learning to further improve already advanced sensor technologies. The fifth generation of the multifunction camera from Continental, which 2020 will go into production, uses not only classic computer vision processes but also neural networks that can be scaled and refined with regard to the intention and gesture classification of pedestrians.
“AI plays a major role in taking on human tasks. With AI software, the vehicle will be able to interpret even complex and unpredictable traffic situations - it is no longer about what is in front of the vehicle, but what could be coming, ”says Karl Haupt, Head of the Driver Assistance Systems division at Continental. “We see AI as a key technology, especially for automated driving. AI is part of the automotive future. ”Accordingly, the performance of AI must be on par with that of a human being. "The car should become so intelligent that it understands both its driver and its environment," says Robert Thiel, Head of Machine Learning in the Driver Assistance Systems division.
Knight Rider in "Real"
AI algorithms are said to resemble an experienced driver so that they can correctly assess and respond to the intent of a pedestrian approaching. But just like humans, such AI systems first have to learn these skills and are almost sent by Continental to the "driving school". In so-called “supervised learning”, the software analyzes a huge amount of data in order to derive promising strategies for action. This knowledge should then be applied later in the vehicle. Of course, the AI is constantly being developed. For this purpose, the driver assistance systems have, for example, suitable data in the form of recorded radar and camera signals from real trips. This technology is to be put into series production as soon as the fifth generation of cameras is introduced. The German company is cooperating with numerous research institutes and has even opened a competence center for deep learning machines this year. By the end of the year, around 400 engineers with special AI know-how will continue to work on the dream of a safe future for people.