Inside Volvo’s Innovative AI-Powered Safety Programme
Volvo has revealed how using AI-generated, 3D worlds is enhancing and improving vehicle safety software at a speed and scale not possible before.
The company is using synthesised incident data collected by advanced sensors in its cars to analyse, reconstruct and explore how incidents can be avoided more effectively.
This approach is enabled by an advanced computational technique called Gaussian splatting.
Underpinned by an AI supercomputing platform from Nvidia, this technique enables Volvo to create vast amounts of realistic, high-fidelity 3D scenes and subjects from real-world visuals.
“We already have millions of data points of moments that never happened that we use to develop our software," says Alwin Bakkenes, Head of Global Software Engineering at Volvo Cars.
"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. This has the potential to unlock a scale that we’ve never had before and even to catch edge cases before they happen in the real world.”
What is Gaussian splatting?
Gaussian splatting is a 3D rendering technique that uses tiny 3D Gaussians (or ‘splats’) to represent a scene.
Instead of using traditional methods like meshes made of triangles, it fills the space with these Gaussians, each having properties like position, size, orientation, colour and transparency.
To make a 3D scene, multiple images of the same scene are taken from different viewpoints. The Gaussian Splatting algorithm then uses these images to optimise the properties of the Gaussians, effectively learning how to represent the scene accurately.
This method creates smooth and detailed images, even when confronted with complex shapes and effects like transparency or reflections.
It makes image rendering fast, often achieving real-time frame rates, and is ideal for applications such as virtual and augmented reality.
Creating safer cars with AI
Volvo uses advanced on-board sensors in its cars to collect essential data such as emergency braking, sharp steering or manual intervention.
Applying this as part of the Gaussian splatting technique allows the company to show a wide range of traffic situations to safety software like advanced driver assistance systems (ADAS), at speed and scale.
The virtual environment created by the technique can be altered or manipulated to change traffic behaviour and conditions, add or remove obstacles on the road and generate different outcomes.
“We now can develop software that works well also in complex, rare yet potentially dangerous ‘edge cases’ and reduce the time it takes to expose our software to edge cases, from months to days,” says Volvo in a press release announcing the development.
Investing in autonomous vehicles
Volvo uses these kinds of virtual environments as part of its wider real-world testing programme for software training, development and validation because they’re safe, scalable and economically sound.
The company creates them in-house through Zenseact, an AI and software company founded in 2017 as a joint venture between Volvo Cars and automotive safety supplier Autoliv.
Zenseact specialises in AI and software development for autonomous driving and ADAS technologies, with the goal of improving car safety and reducing accidents.
It leads the development of advanced driver assistance innovation for Volvo and Polestar cars, and collaborates with global automotive technology provider ECARX.
To use Gaussian splatting, Volvo also relies on the computing power of Nvidia.
The company has a long relationship with the tech giant centred around developing software-defined cars that bring together next generation safety, connectivity, data and software.
Volvo’s new EX90 is built on a centralised core compute architecture from Nvidia capable of over 250 trillions of operations per second (TOPS).
This orchestrates everything in the vehicle, from introducing safe autonomously driving to powering Volvo’s AI-based driving assistance technology.
In a September 2024 press release discussing this collaboration, Volvo explained how it intends to introduce cars built on Nvidia’s DRIVE Thor system, an AI-powered centralised computer for safe and secure autonomous driving.
“Advances in accelerated computing and AI are moving at lightspeed,” said Jensen Huang, founder and CEO of Nvidia, when Nvidia launched the Thor technology in 2022.
“DRIVE Thor is the superhero of centralised compute, with lightning-fast performance to deliver continuously upgradable, safe and secure software-defined supercomputers on wheels.”
Explore the latest edition of AI Magazine and be part of the conversation at our global conference series, Tech & AI LIVE.
Discover all our upcoming events and secure your tickets today.
AI Magazine is a BizClik brand
