Volvo Cars is utilizing Artificial Intelligence (AI) to elevate automotive safety to a new level, using hyper-realistic virtual scenarios to develop automotive safety software more quickly and effectively.
By benefiting from this innovation, Volvo Cars can synthesize real data collected by the sensors in their vehicles, such as emergency braking or sudden changes in direction, to analyze, reconstruct, and anticipate risk situations.
The major revolution lies in the use of the “Gaussian splatting” technique, an advanced computational approach that allows for the creation of ultra-realistic 3D environments from real-world images.
In this way, Volvo can access an unlimited testing space, where it is possible to manipulate variables – adding or removing road users, changing traffic behavior, or creating unexpected obstacles. This means that Volvo Cars’ safety software can be trained and validated in days instead of months, enabling the simulation of even the most extreme and rare scenarios, those that can make all the difference in the real world.
“We already have millions of data points from moments that never happened, which we use to develop our software. Thanks to ‘Gaussian splatting,’ we can take one of these rare cases and replicate it into thousands of new variations of the scenario to train and validate our models. This has the potential to unlock a scale we have never had before and even to identify extreme cases before they occur in the real world.”, said Alwin Bakkenes, Head of Global Software Engineering at Volvo Cars.
