Photo credit: www.therobotreport.com
Torc Robotics Inc. has recently revealed its collaboration with Flextronics International Ltd. and NVIDIA Corp. to create a scalable artificial intelligence compute system specifically designed for autonomous trucks. The company showcased its advanced technology at the NVIDIA GTC event held in San Jose, California, last week.
“We were thrilled to debut at GTC while announcing our partnership with NVIDIA and Flex,” stated CJ King, the chief technology officer of Torc, during the event. “As we approach commercialization, it’s crucial for us to showcase our cutting-edge technology and engage with our talented engineers, spreading the word about Torc.”
Established in 2005, Torc functions as an independent subsidiary of Daimler Truck AG. Based in Blacksburg, Virginia, the company focuses on developing self-driving technology, aiming to bring autonomous trucks to the U.S. market specifically for long-haul purposes.
Torc Emphasizes Physical AI
Torc Robotics highlighted that its physical AI technology empowers self-driving trucks to accurately perceive, comprehend, and execute complex tasks in the physical environment. The company claims this physical AI framework is integral to its software, enabling vehicles to navigate autonomously by leveraging various sensors including lidar, radar, and cameras.
This technology equips the trucks to make critical decisions regarding lane changes, braking, and obstacle avoidance, fostering safe and efficient driving experiences. Torc asserts that they are pioneering the application of physical AI in production-scale autonomous long-haul trucking.
Besides its offices in Austin, Texas, and Montreal, Quebec, Torc has established a fleet operations facility in the Dallas-Fort Worth area and has a presence in Ann Arbor, Michigan.
NVIDIA DRIVE AGX as Compute Backbone
During the NVIDIA GTC event, Torc disclosed its close collaboration with NVIDIA to develop a multi-chip adaptable architecture featuring the DRIVE AGX platform, which incorporates the DRIVE Orin system-on-a-chip (SoC) and the DriveOS operating system.
“NVIDIA DRIVE AGX has a proven track record in delivering real-time automotive applications at the edge,” remarked Rishi Dhall, NVIDIA’s vice president of automotive. “It offers the significant compute performance, minimal latency, and multi-sensor connectivity crucial for the sophisticated autonomous trucking software developed by Torc, ensuring robust perception, prediction, and planning for safe operations.”
Dhall further noted that Torc is poised for scalable production with a target for commercial rollout in 2027 while seamlessly transitioning to NVIDIA DRIVE AGX equipped with DRIVE Thor.
The collaborative platform utilizes NVIDIA DRIVE AGX alongside Flex’s Jupiter compute platform, aimed at facilitating product development and market entry by 2027.
Flex’s Contribution to Software-Defined Vehicles
Flextronics is supplying its Jupiter compute design alongside its advanced manufacturing capabilities, which will enable Torc to create high-performance production hardware and software adapted for an autonomous-ready Freightliner Cascadia. This model includes advanced technologies and redundant systems to facilitate the deployment of autonomous driving features in the future.
“Our partnership with Torc, Daimler Truck, and NVIDIA demonstrates how Flex collaborates across the ecosystem to assist mobility firms in launching next-gen technology effectively and quickly,” said Mike Thoeny, president of automotive at Flex. “We value the confidence that Torc and Daimler Truck have in us as we apply our Jupiter compute platform and manufacturing expertise to enable large-scale autonomous long-haul trucking.”
Torc indicated that the NVIDIA DRIVE technology, coupled with Flextronics’ automotive compute designs, meets its demanding standards for size, performance, cost, and reliability. This solution also aligns with the total cost of ownership (TCO) expectations of Torc’s fleet customers who are interested in integrating nonstop, driverless long-haul trucking.
“This effort exemplifies what it means to have a true software-defined vehicle (SDV),” the company stated. “It is equipped to adapt to constantly evolving operational design domains, including new lanes, routes, hubs, hardware and sensor configurations, operational regulations, and varying road conditions.”
Torc Robotics has previously validated its driverless product acceptance through a successful testing platform at highway speeds on a closed course last year.
Source
www.therobotreport.com