Where VR Meets the Road: How GPUs Power ‘Hack Rod’, World’s First AI-Generated Car
The team at Hack Rod aims to create the world’s first car engineered with artificial intelligence and designed in a virtual environment — and may well reinvent the manufacturing supply chain in the process.
And because the Los Angeles-based startup will need a ton of computational horsepower they turned to design software leader Autodesk and NVIDIA GPU technology.
Computers that use AI to creatively come up with design ideas on their own are at the heart of what’s known as generative design. Driving the trend: advancements in AI and powerful GPUs that can quickly simulate complex phenomena. These have enabled software to play an active, participatory role in the invention of form.
Generative Design with Autodesk Dreamcatcher
Autodesk has adopted this concept with Dreamcatcher. Share your goal with the computer, tell it what you want to achieve, as well as the constraints involved, and the software generates ideas that expand on the creative possibilities for a designer.
The Hack Rod team fabricated a chassis with proven geometries but the data captured was made up of physical forces affecting the car and driver. During several test drives, the car and driver were rigged with hundreds of sensors to measure stresses, etc., and that data was fed into Autodesk Dreamcatcher to generate new chassis designs.
Then, they fed it to Dreamcatcher software, which uses NVIDIA GPUs to deliver the computational horsepower necessary necessary to quickly analyze and produce the resulting recommendations.
Using Autodesk VRED virtual prototyping tools, the resulting recommendations will be shown as part of a VR collaborative design review, powered by Silverdraft Demon VR, taking place in NVIDIA booth 509 at SIGGRAPH, running July 24-28 in Anaheim, Calif.
“The Democratization of Manufacturing”
“The way your world opens up from a design perspective in photoreal VR is just magic,” says Mouse McCoy, Hack Rod founder, creative director and former professional race car driver. “The speed at which you can make decisions about your final product is unrivaled and when you start to add in AI/machine learning, it’s like you have 1,000 engineers working for you solving problems in a fraction of the time that it used to take. It’s the democratization of manufacturing.”
Once a final design is selected, it’s handed off to Autodesk Design Graph, a machine learning search application that makes parts recommendations (think actual nuts and bolts) to match the criteria.
Generative design is creating complex engineering results that were previously not able to be manufactured. The rapid progress in the advanced manufacturing space is now allowing for this complexity to be produced. Hack Rod, Autodesk and NVIDIA will showcase some large metal 3D printed examples of this at Autodesk University in November, in Las Vegas, and we’ll be showing a full car going from Game to Garage at SXSW in March, in Austin, Texas.
“The future of making things looks really cool,” says McCoy. “When you combine collaborative photo-real 3D VR design with AI-based generative design, machine learning and advanced manufacturing, it creates a supply chain of the future that puts the power of large organizations in the hands of the Everyman.”
Learn more by checking out the Autodesk Design Graph and VRED talks at SIGGRAPH. See the complete listing of GPU computing talks here. And follow the latest at the show at #SIGGRAPH2016.