GM, Gatik, and Torc Partner with Nvidia to Boost Self-Driving Innovation

The NVIDIA GTC conference has commenced with a series of announcements showcasing its pivotal role in promoting advancements in autonomous driving technology.

NVIDIA provides automakers and autonomous vehicle firms with a suite of tools to enhance self-driving cars and aid in creating digital twins of factories. On Tuesday, companies such as General Motors, Gatik, and Torc disclosed their plans to utilize Nvidia products for vehicle manufacturing and automated driving technology.

The wide array of Nvidia’s hardware and software solutions can be somewhat confusing, so here’s a quick guide to their terminology:

• Drive AGX: A powerful in-vehicle supercomputer that processes real-time sensor information.
• Drive Orin SoC: An advanced system-on-a-chip that processes data from various sensors and supports new vehicles with Level 4 autonomous capabilities.
• Drive Thor SoC: This system is optimized for Transformer architecture, enabling it to handle self-driving tasks, cockpit functionalities, and infotainment systems.
• DriveOS: A safety-centric operating system for Nvidia’s autonomous vehicle platform, ensuring real-time AI processing and integration of advanced driving features.
• Omniverse: A simulation platform allowing automakers to create intricate, AI-driven virtual environments for testing AV software and building digital factory replicas.
• Cosmos: Designed to enhance world model training for physical AI applications in autonomous vehicles and robotics.

Additionally, Nvidia has introduced Halos, an AI-enhanced safety system for autonomous vehicles and robots. This system integrates various Nvidia automotive hardware and software safety features into a comprehensive solution.

General Motors

General Motors has broadened its partnership with Nvidia, impacting all areas of its operations, including factories, robots, and self-driving vehicles. Regarding production facilities, GM plans to utilize Omniverse in conjunction with Cosmos to train AI manufacturing models. This will aid in constructing cutting-edge factories. Omniverse will facilitate the creation of digital twins of their factories, allowing GM to test new production methods virtually without disrupting current operations. Moreover, the platform will assist in training robotic systems for tasks such as material handling and transportation.

For its self-driving vehicles, GM will employ Nvidia's Drive AGX as the in-vehicle technology for future advanced driver-assistance systems and in-cabin safety enhancements.

Gatik

Self-driving truck company Gatik, which has support from Isuzu and Goodyear Ventures, is joining Nvidia’s automotive ecosystem. This company focuses on autonomous logistics with their self-driving box trucks. They will develop and implement Drive AGX, powered by the Drive Thor, to serve as the AI manager for their fleet. Gatik will also utilize DriveOS for running safety-focused AI models.

The partnership aims to speed up the deployment of Level 4 autonomous trucks for clients like Walmart, Kroger, and Tyson Foods.

Plus

Another player in the autonomous trucking arena, Plus, announced its intention to leverage Cosmos world foundation models to enhance the testing and development of its SuperDrive autonomous driver system.

Built on the Nvidia Drive AGX platform, the SuperDrive is part of Plus's initiatives in developing AV 2.0 technologies, including generative AI and visual language models. As noted earlier, the AGX platform is typically suited for advanced driver-assistance functionalities. For more complex autonomy requirements, companies often turn to Nvidia's Orin or Thor SoCs.

Plus has recently formed partnerships with commercial vehicle makers like Traton Group, IVECO, and Hyundai to integrate SuperDrive into their products. The startup is currently testing its technology on public roads in both Texas and Sweden, with a view toward commercial launch by 2027.

Torc

Torc, another autonomous trucking company, is also aligning with Nvidia to create an efficient physical AI compute system for its AV technology. Based in Virginia and a subsidiary of Daimler Truck AG, Torc has partnered with Flex to develop automotive-grade computing solutions.

The company is utilizing multiple Nvidia chip architectures, including Drive AGX, Drive Orin, and DriveOS, to implement future autonomous driving capacities, targeting a commercial roll-out by 2027. Recently, in October 2024, Torc completed its inaugural testing with no driver present on a closed course in Texas.

Volvo

In contrast, while Volvo is not collaborating with Nvidia for advancing automated driving technology, it is utilizing Nvidia’s Blackwell GPUs for aerodynamics simulations. Rather than adopting Nvidia's Omniverse simulator, Volvo is collaborating with Ansys, a simulation software firm. Ansys’s Fluent simulation tool, powered by eight Blackwell GPUs, has assisted Volvo in designing the new EX90 electric vehicle by reducing aerodynamic drag, subsequently enhancing battery efficiency.

Ansys reports that its Fluent simulator has significantly decreased the simulation runtime from 24 hours to just 6.5 hours, enabling more design iterations daily, optimizing vehicle designs, and hastening time to market.

technology, automotive, innovation