Volvo is utilizing AI-generated, life-like virtual worlds to advance the development of its safety software, including driver assistance systems (ADAS), with the goal of making its cars even safer.
The Swedish automaker can now synthesize incident data collected by the advanced sensors in its new cars, such as emergency braking, sharp steering, or manual intervention. This enables them to probe, reconstruct, and explore these incidents in new ways to understand better how they can be avoided.
This is made possible by an advanced computational technique called Gaussian Splatting, which can generate many realistic, high-fidelity 3D scenes and subjects from real-world visuals. The virtual environment can be manipulated by adding or removing road users and altering the behavior of traffic or obstacles on the road to create different outcomes.
Related
How Nvidia’s Cosmos Platform Could Change Transportation
Nvidia’s Cosmos platform could revolutionize transportation with AI-driven simulations, optimizing traffic, autonomous tech, and urban planning.
Such a technique allows Volvo to expose its safety software to all types of traffic situations at a speed and scale previously unattainable. The automaker can now develop software that performs effectively in complex, rare, yet potentially dangerous “edge cases,” reducing the time it takes to expose their software to these scenarios from months to days.
“We already have millions of data points of moments that never happened that we use to develop our software. Thanks to Gaussian Splatting we can select one of the rare corner cases and explode it into thousands of new variations of the scenario to train and validate our models against.”
– Alwin Bakkenes, Volvo’s Head of Global Software Engineering
One Part of the Puzzle
Volvo utilizes virtual environments alongside real-world testing for software training, development, and validation due to their safety, scalability, and cost-efficiency. These virtual environments are developed in-house in collaboration with Zenseact, an AI and software company founded by Volvo.
This project is part of a PhD program with leading Swedish universities to explore the potential integration of neural rendering techniques into future safety initiatives. The Wallenberg AI, Autonomous Systems, and Software Program (WASP) sponsored the study.
A History of Using Data to Improve Safety
Volvo has a long-standing history of using data and advanced technologies to enhance safety. Data collected by the Volvo Cars Safety Research team has been instrumental in developing and testing some of the world’s most prominent safety features.
In the 1970s, Volvo began leveraging data to improve safety, thanks to its Safety Research team. In the early days, the team would arrive at accident scenes with measuring tapes, evaluating skid marks and other crash indicators.
The data and insights gathered from these accidents inspired numerous life-saving innovations, such as the Whiplash Injury Protection System and the Side Impact Protection System. Today, advanced technology allows them to use data even more effectively to prevent risky situations.
Integration of NVIDIA Technology
Thanks to its recently expanded partnership with NVIDIA, Volvo can explore technologies like Gaussian Splatting. The new generation of fully electric cars, built on NVIDIA-accelerated computing, collects data from various sensors to better understand what’s happening in and around the car.
An AI supercomputing platform powered by NVIDIA DGX systems contextualizes this data, unlocks new insights, and trains future safety models, enhancing and accelerating the development of artificial intelligence. This supercomputing platform is part of a recent investment by Volvo and Zenseact to establish one of the largest data centers in the Nordics.
Source: Volvo Cars
