As physical AI-powered edge systems and infrastructure increasingly automate, they must autonomously perceive, plan, and execute complex tasks—from traffic pattern detection and industrial inspection to autonomous mobile robots in warehouses and logistics.
To develop and deploy the next generation of autonomous AI systems, a new framework is required. This involves training multimodal, generalized AI models for various tasks, then testing and validating these models and their associated software in simulation. Finally, the entire stack is deployed on the physical edge AI system to perform actions in real time.
NVIDIA’s three computers—for training, simulation, and deployment—are essential for achieving human-like intelligence for autonomous edge solutions.
NVIDIA DGX is the optimal platform for training large, generalized AI foundation models to perceive and act with the physical world. It delivers the optimal mix of compute, networking, and GPU-optimized frameworks, like NVIDIA NeMo™, deployable using NVIDIA NIM™.
NVIDIA OVX enables synthetic data generation, robot learning with NVIDIA Isaac™ Lab, and testing with OpenUSD and Omniverse Cloud Sensor RTX™ assess various scenarios using physically accurate simulations.
NVIDIA Jetson™ and IGX give you scalable platforms to deploy your entire robot stack with powerful AI compute, high-speed I/O, and NVIDIA AI software, including NVIDIA Metropolis for vision AI applications.