NVIDIA의 AI 훈련, 시뮬레이션 및 차량 내 컴퓨팅을 위한 모듈식 3-컴퓨터 플랫폼은 기초 기술
The global robotaxi market is projected to reach 400 billion dollars by 2035, with over 6 million commercial vehicles in operation as driverless fleets already move people through some of the world's busiest and most complex streets. Deploying a driverless vehicle presents one challenge; scaling a fleet to thousands of vehicles represents a next-level computing challenge that demands delivering the same safe, reliable performance across all of them.
Meeting these demands requires enormous amounts of compute across the robotaxi development lifecycle, from preparing and training AI models to simulating and validating driving behavior, as well as real-time processing in the vehicle. NVIDIA provides an open platform for AI training, simulation and safety validation, with libraries, software development kits, workflows and models that developers can use alongside their own technology stacks. Every major robotaxi program operating at commercial scale today runs on NVIDIA's modular stack, spanning AI training, simulation, in-vehicle computing or a combination of the three.
A robotaxi technology stack is the end-to-end set of technologies used to develop, validate and deploy autonomous vehicles, from data and AI model training to simulation, safety validation and real-time in-vehicle computing. NVIDIA's robotaxi and autonomous vehicle platform brings these capabilities together in a three-computer solution: the training computer, simulation and validation computer, and in-vehicle computer.
The training computer uses NVIDIA DGX systems, where robotaxi intelligence advances as programs turn growing volumes of fleet data into increasingly capable models. NVIDIA's Alpamayo portfolio of open reasoning vision language action models, simulation frameworks and physical AI datasets gives developers building blocks they can adapt to their own data, requirements and technology stacks. On a challenging autonomous driving evaluation, adding meta-action and chain-of-thought reasoning data improved a vision language action model's trajectory prediction accuracy, reducing minimum average displacement error by 43 percent, from 2.08 to 1.18.
The simulation and validation computer runs on NVIDIA Omniverse and Cosmos on NVIDIA RTX PRO servers. NVIDIA Omniverse NuRec models reconstruct real-world driving scenarios from sensor data, while NVIDIA Cosmos world foundation models generate physically based variations of them, enabling developers to turn thousands of real-world corner cases into millions of combinations of driving behavior, traffic, weather, lighting and sensor conditions. The NVIDIA AlpaSim simulation framework extends the workflow for training and evaluating reasoning-based autonomous driving models, helping developers identify weaknesses before deployment.
The in-vehicle computer uses NVIDIA DRIVE Hyperion, NVIDIA's modular in-vehicle compute and sensor reference architecture for level-4-ready robotaxis. DRIVE Hyperion 10 pairs dual NVIDIA DRIVE AGX Thor systems-on-a-chip, built on the NVIDIA Blackwell platform, with 14 high-definition cameras, nine radars, three lidars and 12 ultrasonics for real-time 360-degree sensor fusion. Its redundant compute and sensing design supports fail-operational driving if a sensor or compute component fails.
NVIDIA's robotaxi ecosystem spans every region where commercial robotaxi services are emerging today: Asia, Europe, the Middle East and North America. Uber is scaling its fleet of NVIDIA DRIVE Hyperion with plans to reach 28 cities by 2028. Lyft plans to use NVIDIA DRIVE Hyperion as a reference architecture for future autonomous fleets. Waymo partners with NVIDIA to help build its autonomous computing system. WeRide plans to bring its DRIVE Hyperion and DRIVE AGX Thor-based vehicle to key markets across Southeast Asia through its partnership with Grab. Other partners adopting NVIDIA's platform include Wayve, Nissan, Autobrains, Zoox, Momenta, Pony.ai, Tensor, Waabi, TIER IV, Isuzu, Lenovo and DeepRoute.ai, among others, with Wayve, Nissan and Uber developing a global robotaxi program using a prototype vehicle that combines Nissan's vehicle engineering, Wayve's embodied AI and the NVIDIA DRIVE Hyperion platform.