20180611-adas-ai-img-pedestriandetection-300x169

Image: Continental

Continental has shown and outlined a computer vision platform for automated vehicles at CES Asia. This uses artificial intelligence, neural networks and machine learning to improve the ‘scene understanding’ and analysis of data from sensors and cameras.

Neural networks will be used in Continental’s next-generation cameras, to go into production in 2020, and the company says that this is part of its vision for an automated vehicle to have decision-making capabilities ‘at least on a par with those of humans.’

“AI plays a large role in taking over human task," Karl Haupt, head of Continental’s ADAS business unit, said in a statement. "With AI software, the vehicle is able to interpret even complex and unforeseeable traffic situations – it’s no longer about what’s in front of me but about what could be in front of me. We see AI as a key technology for automated driving. AI is a part of an automotive future.”

Continental draws a parallel with human driving schools in terms of ‘supervised learning’, whereby the software has to build up knowledge through testing strategies for action. In this ‘virtual driving school’, a huge database of real-life situations is being created to continually inform the algorithms. The fifth-generation cameras will use a system in which the data input – such as the pixels from a camera image – is combined and parametrized, and the system then searched for the combination of parameters to solve the problem.

Continental says that it will be employing around 400 engineers with special AI expertise by the end of 2018, and that it is recruiting people to work in product and process development.

-Farah Alkhalisi