Torc’s autonomous driving system equipped on a Freightliner Cascadia tractor at an industry show. (Seth Clevenger/Transport Topics)
Automated truck developer Torc has partnered with manufacturing company Flex and artificial intelligence specialist Nvidia to develop a scalable physical AI compute system for autonomous trucks, the company announced March 18.
Blacksburg, Va.-based Torc has been working toward autonomous truck production at scale as an independent subsidiary of Daimler Truck. Nvidia worked closely with the company to develop a multi-chip adaptable architecture that leverages Drive AGX, Nvidia’s autonomous vehicle development platform. Flex is providing its Jupiter compute design platform and advanced manufacturing capabilities.
“We view this as a really important next step and milestone,” said Andrew Culhane, chief commercial officer at Torc. “It’s been in the works for quite a while — really leading to that productization phase and getting to that true scalable driver-out launch in [2027].” He added, “We’ve always tried to be very consistent and clear with our plans for development and truly what it takes to launch a product in this space. So we view this as a really exciting announcement toward that.”
Torc said the approach allowed it to develop a scalable and high-performance production platform for Freightliner Cascadia trucks. The platform was validated by internal driverless product acceptance tests on a closed course at highway speeds last year.
“At the end of the day, there are many pieces that are absolutely critical to launching a driver-out product,” Culhane said. “But the core of that really is that computer — that brain — and bringing together Torc software, the Nvidia chipset and then Flex’s design and manufacturing capability. We think that is an amazing foundation to build that scalable driver-out launch on.”
Torc noted the components of this partnership meet its standards for size, performance, cost and reliability, including total cost of ownership targets for fleets looking into nonstop and longhaul driverless trucking integrations. Longer term, Torc said the partnership is adaptable to changing operational design domains.
“What it was all about was finding those right partners,” Culhane said. “Obviously, Nvidia is world class at what they do. Then, on the Flex side, finding somebody that could really meet the true demand — not just the prototype designs, but the production, the safety that’s critically needed — to really set us up for success to launch. That’s where we landed with Flex, and it’s been a great partnership between the three groups.”
The Nvidia Drive AGX was built around autonomous driving applications, with a focus on high compute performance, low latency and multi-sensor connectivity that the Flex platform will support. Torc is confident that leveraging these software solutions will ensure a low-risk and high-confidence path to production.
“That really is that backbone, that core, of our Level 4 self-driving truck,” Culhane said. “That multi-core all-in solution gives not only all the performance we need for automated driving but all the redundancy and safety that are needed. That’s what that foundational layer of that compute gives us.”
RELATED: Aurora Partners With Nvidia, Continental on Driverless Trucks
Culhane noted that these components will be installed on vehicles on the factory line, a process that he said will help the trucks become scalable.
“Scalable for us isn’t 10 or 20 trucks, hand-built running on server-grade components,” Culhane said. “This is truly running on automotive grade with a level of safety, reliability and ruggedness — and ruggedness being the last part because trucking is tough — that we really need to have this product out there.”
Torc plans to showcase the capabilities of the joint solution on its demo truck at Nvidia’s GTC AI conference, which began March 17 and runs through March 21.
