Continental to Work With Berkeley on AI Research
BDD works with state-of-the-art technologies for machine seeing and learning in automotive applications. The multidisciplinary center is managed by the Institute of Transportation Studies at the University of California, Berkeley and industry sponsors support the program to help bring new technologies to automotive applications.
In the first year of the program membership, Continental and BDD are focused on two fields of research. First is the testability of AI algorithms in safety-relevant systems. Drivers need to be sure that the complex technology in their vehicles will work properly, so BDD is developing methods that will allow the reliability of AI systems to be tested more efficiently. The researchers at the centre are also looking at how to operate AI applications in a memory-efficient way to accelerate and optimise neural networks. This will allow easier implementation of AI methods in vehicles.
“We are joining forces with the world’s leading AI researchers,” said Demetrio Aiello, Head of Continental’s Corporate Artificial Intelligence and Robotics Lab. “Building on the momentum of our strategic partnerships with the University of Oxford, DFKI (German Research Center for Artificial Intelligence) and other AI thought leaders, we have signed a five-year agreement to be members of the UC Berkeley DeepDrive (BDD) center.”
“What inspired us most to team up with the experts in Silicon Valley and UC Berkeley was the highly interesting research in the field of Explainable AI as well as the optimisation of deep neural networks that were taking place there,” Aiello added.
Explainable AI focuses on understanding precisely how an AI system makes decisions. To test artificial intelligence in detail, experts must know exactly how it works. In addition to the benefits of the research itself, Aiello shared another important advantage of the membership.
“The opportunity to have colleagues from Silicon Valley and other Continental locations working as part of BDD research teams enables more efficient collaboration and transfer of expertise. It also allows us to identify the talent we need for our AI strategy at an early stage.”
07 Sep 2018