TNO: Self-driving car can predict traffic situations in seconds with the help of AI

TNO has successfully tested a self-driving car that can predict what might happen within seconds using artificial intelligence. The research organization reported this in its annual report published on Friday.

"A few seconds may not seem like much," said TNO, "but those seconds can save lives in dangerous traffic situations. Because the driver of a vehicle still has the opportunity to take control safely."

Experiments in San Francisco showed that self-driving cars still have major difficulties in unknown situations. With the new system, the AI in the TNO car can predict what behavior to expect in the lane. A trial with a concept version of the car was successful on a test track. A TNO spokesperson assumes the new system can be introduced "in the foreseeable future."

TNO is also researching the use of AI for a "moral compass" in robots that have to rescue people before the rescuers arrive on the scene. In doing so, the robots learn to make their own trade-offs, such as whether a door should be opened in a particular situation. "For AI systems to be able to judge situations based on our own behavior, they need to be based on a solid framework that we consider morally right."

According to TNO, there is also a strict technology framework for automated vehicle systems. “You then have to start fitting those vehicles into the national traffic systems, and they are different in every country. We think it’s important to also look at typical Dutch traffic situations, so things like bicycle traffic, pedestrians, roundabouts, rush-hour lanes, and so on,” Senior Consultant in Vehicle Safety Assessment Methodologies Olaf op den Camp said.

TNO managed to increase its revenue by 16 percent to 688 million euros last year. The increase came from both government and industrial projects.