Continental taps Berkeley University’s AI expertise

July 30, 2018 // By Christoph Hammerschmidt
The importance of artificial intelligence (AI) for the automation of driving apparently cannot be overestimated. The vehicle supplier Continental has now announced a further research cooperation in this field. But there is still a considerable need for research: In the current case it is a matter of optimising computational processes and safety.

Continental is expanding its international artificial intelligence network to Silicon Valley. Following similar agreements with the University of Oxford and the German Research Institute for Artificial Intelligence (DFKI), the Hanover-based automotive supplier has now signed a five-year contract with the Berkeley DeepDrive (BDD) research group at the University of California.

Among other things, the joint work focuses on optimizing the speed of neural networks in the car and securing AI systems in safety-critical applications. The main goal of the cooperation: AI research results should find their way into series production as quickly as possible.

Berkeley DeepDrive deals with state-of-the-art technologies for machine vision and learning in automotive applications. Home of the multidisciplinary center is the Institute of Transportation Studies at the University of California at Berkeley. Sponsors from industry promote cooperation between different disciplines in order to bring new technologies into automotive applications. Berkeley DeepDrive's Director is Professor Trevor Darrell, Head of the Partners for Advanced Transportation Technology (PATH) Department.

Two research areas of the Berkeley DeepDrive are in the foreground for Continental in the first year of the partnership. One is the "testability of AI algorithms in safety-relevant systems": drivers must be able to rely on the fact that the complex technology of their vehicles functions reliably in operation. That is why BDD develops methods with which the reliability of AI systems can be better tested. On the other hand, Berkeley researchers are investigating ways to make AI applications particularly memory efficient in order to accelerate and optimize neuronal networks. This makes it easier to use the AI methods in vehicles later on.

"Our involvement in Silicon Valley and Berkeley is supported above all by the highly interesting research in the discipline of Explainable AI, i.e. explainable artificial intelligence, and the optimization and compression of deep neural networks that take place there," explains Demetrio Aiello who overlooks Advanced AI Research for Continental.


Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.