Volvo Leverages AI and Virtual Worlds to Advance Car Safety

Volvo Cars uses AI and Gaussian splatting to create virtual worlds for ADAS testing, enhancing safety and accelerating edge case simulation. Learn more.

Volvo Cars is taking vehicle safety development to the next level by using generative AI and virtual worlds. Leveraging Gaussian splatting technology, the company creates highly realistic 3D scenes to simulate complex traffic scenarios and train its advanced driver assistance systems (ADAS).

According to Alwin Bakkenes, Head of Global Software Engineering at Volvo Cars, this technology enables the reproduction of rare and potentially dangerous edge cases in thousands of variations, reducing testing time from months to days. It opens new possibilities for preparing software to handle challenging situations on the road.

The virtual environments are developed in-house in collaboration with Zenseact and are part of a PhD program with leading Swedish universities, supported by the Wallenberg AI, Autonomous Systems and Software Program (WASP). The project explores the integration of neural rendering techniques into future safety initiatives.

A key role is also played by Volvo’s strategic partnership with NVIDIA. Using NVIDIA DGX supercomputers, Volvo accelerates data analysis from vehicle sensors and improves AI model training efficiency. These technologies will form the foundation of software development for Volvo’s next-generation electric cars.

Historically, Volvo was among the first carmakers to systematically use data for safety improvements. Today, the company reaffirms its leadership by integrating cutting-edge AI methods. These developments were presented by Volvo Cars at the NVIDIA GTC conference.

Source: volvocars.com