The widespread evolution and adoption of AI has not escaped the attention of Volvo. The Swedish car maker is set to redefine automotive safety by harnessing the power of AI-generated, lifelike virtual environments to develop cutting-edge driver-assistance systems (ADAS).
In a groundbreaking initiative, the company is utilising advanced computational techniques to create virtual simulations that enhance the development and validation of its safety software, significantly improving the potential to prevent accidents.
Advancing Safety Through Tech
The key technology behind this innovation is Gaussian splatting, a sophisticated method that creates highly realistic 3D virtual scenes based on real-world visual data. By using this technique, Volvo is able to reconstruct and manipulate traffic scenarios that would be impossible or dangerous to simulate in real life.
For instance, the virtual world can be adjusted to add or remove road users, change traffic behaviour, or simulate different obstacle scenarios, allowing for more accurate and efficient testing of safety systems.
This approach is transforming the way Volvo develops its safety software. By generating variations of rare and complex edge cases—those potentially dangerous, yet unlikely, scenarios—the car maker can now test its systems at a scale and speed previously unattainable.
What once took months of real-world testing can now be done in just days, dramatically reducing development time while improving system reliability.
Virtual To Reality
Volvo Cars uses AI and virtual worlds with the aim to create safer cars.
The integration of these virtual environments complements traditional real-world testing, providing a safer, more scalable, and cost-effective solution for software development and validation. These virtual environments, developed in partnership with Zenseact, an AI and software company founded by Volvo Cars, are part of a broader initiative to enhance vehicle safety.
Volvo’s commitment to using data to improve safety is nothing new. Since the 1970s, the company has led the way in utilizing accident data to develop life-saving innovations. Early efforts by the Volvo Cars Safety Research Team involved collecting real-world data from crash sites, measuring skid marks, and identifying key factors that could prevent injuries. These findings led to the creation of safety features such as the Whiplash Injury Protection System and Side Impact Protection System.
Today, Volvo continues to use data more intelligently, leveraging new technologies to predict and prevent dangerous situations. Through a strengthened partnership with NVIDIA, Volvo has integrated AI-powered supercomputing into its safety development process.
The data collected by its sensors is now contextualized and processed using NVIDIA DGX systems, enabling faster and more accurate insights that fuel the development of AI-driven safety models. These efforts represent a new era of safety—where AI, data, and virtual environments work together to predict, prevent, and ultimately save lives on the road.
