公告
Flex 推出模块化机架级冷却液分配装置，扩展数据中心冷却产品组合
Flex公司
联系我们
Flex 资源 Torc collaborates with Flex on physical AI platform for autonomous trucks

Torc collaborates with Flex on physical AI platform for autonomous trucks

发表于
1 10 月, 2025
实例探究 汽车

Advanced Driver Assistance Systems (ADAS) and fully autonomous solutions hold immense potential for 下一代移动出行.

Flex Jupiter Compute Platform

Flex developed an NVIDIA-based hardware platform that will support Torc’s productization and scaled market entry for autonomous trucks.

The Flex Jupiter Compute Platform enables Torc and automotive original equipment manufacturers (OEMs) to deploy advanced physical AI at scale on a robust, cost-effective, and flexible hardware foundation.

下载案例研究

Flex Jupiter Compute Platform
Production ready

挑战

To address the complexity of ADAS and advance toward full autonomy in commercial and passenger vehicles alike, computing hardware must be powerful enough to handle demanding software models, cost-effective enough to be deployed at scale, and flexible enough to be adopted across a range of programs and OEMs. This requires seamless scalability to meet high-performance computing demands. To advance its L4 self-driving semi-trucks, Torc engaged Flex to accelerate the launch and deployment of next-gen, autonomous long-haul trucking technology.

Torc Daimler Truck
德州仪器汽车系统总监 Mark Ng 和 Flex 汽车工程副总裁 Andreas Heim
视频
探索电动汽车系统中碳化硅的电流传感解决方案
汽车备用DC/DC转换器
白皮书
备用 DC/DC 转换器
Mike Theony 接受 PACE 奖项
新闻
庆祝PACE汽车创新奖

解决方案

In support of Torc’s autonomous trucking ambitions, Flex used the latest NVIDIA DRIVE AGX technologies and an innovative, highly scalable approach to develop a unique solution that speeds time to market and reduces costs without sacrificing safety or quality.

  • Each Jupiter unit features two NVIDIA DRIVE AGX Orin System-on-a-Chips
  • Up to four Jupiter units can be connected for networked parallel computing, providing computing power sufficient for L4 autonomous driving
  • Four units provides 2,000 TOPS of compute, 64 vision sensors, and 112 networking I/O — plus Automotive Audio Bus (A2B) for commercial vehicle applications
  • Scalability helps eliminate the need to split codebases for different ADAS levels, hardware generations, or software implementations

结果

Front of a Torc Daimler truck

As the first ADAS and autonomous driving computing platform with networked parallel computing, Jupiter sets a new standard for scalable automotive hardware.

The customizable, safety-focused, and market-ready platform enables Torc to deliver autonomous long-haul trucking at scale.

  • Jupiter units in use on public roads for Torc’s Virtual Driver testing, on path to full product launch
  • A hardware platform for purpose-designed, cost-competitive, and high-performance computing that supports strict OEM safety cases
  • Accelerated validation through software-agnostic approach with published software interface and functional safety documentation

The Jupiter Compute Platform provides Torc with the performance and production-intent reliability needed to deploy our AV 3.0 autonomous driving software today. We see this as an important factor that helps us deliver a safe and profitable product at scale.

— CJ King, Chief Technology Officer, Torc Robotics