Listen to this article

NVIDIA’s founder and CEO Jensen Huang unveiled a number of new projects from the company in his first in-person keynote in almost four years. NVIDIA’s latest innovations aim to ride the wave of generative AI advancements made in recent months. 

“We’re now at the tipping point of a new computing era with accelerated computing and AI that’s been embraced by almost every computing and cloud company in the world,” Huang said at the conference. Huang also noted that 40,000 large companies and 15,000 startups now use NVIDIA technologies with 25 million downloads of CUDA software last year alone.

Metropolis for Factories

NVIDIA Metropolis for Factories is made up of a variety of factory automation workflows that enable industrial technology companies and manufacturers to develop, deploy and manage customized quality control systems. 

Metropolis now includes a state-of-the-art AI platform and workflows to develop incredibly accurate inspection applications, like automated optical inspection (AOI). One of NVIDIA’s manufacturing customers, Pegatron, is using Metropolis for Factories for AOI on its production line. 

Pegatron manufactures a variety of items, from motherboards to smartphones, laptops, and gaming consoles. The company is using the entire Metropolis for Factories workflow to support its printed circuit board (PCB) factories with simulation, robotics, and automated product inspection. With Metropolis for Facotries, Pegatron has achieved 99.8% accuracy on its AOI systems. 

Huang announced that more than 50 manufacturers and industrial automation providers, including Foxconn, Industrial Internet, Pegatron, Quanta, Siemens, and Wistron, are implementing Metropolis for factories. 

Isaac AMR

Isaac AMR, which aims to bring advanced mapping, autonomy and simulation to mobile robots, will soon be available for early customers, according to Huang.

Developers can use Isaac AMR to simulate, validate, deploy, optimize, and manage fleets of autonomous mobile robots (AMRs). The platform also includes edge-to-cloud software services, computing, and a set of reference sensors and robot hardware to accelerate development. 

Isaac AMR speeds up the deployment process from weeks to days by tying semantic understanding of large environments with DeepMap’s cloud-based service. It offers centimeter-level accuracy and generates rich 3D voxel maps, which can be used to make occupancy maps and semantic maps for multiple types of AMRs.

The platform is built on the foundations of NVIDIA’s Nova Orin reference architecture, which serves as the brains and eyes of Isaac AMR. Nova Orin integrates multiple sensors, including stereo cameras, fisheye cameras, and 2D and 3D lidars, with NVIDIA’s Jetson AGX Orin system-on-modules. 

Isaac AMR can also simplify robot operations by tapping into physics-based simulation from Isaac Sim, powered by NVIDIA Omniverse, which is an open development platform for industrial digitalization. Isaac Sim can bring digital twins to life, so the robot application can be developed, tested, and customized for each customer before being deployed in the physical world. 

Jetson AGX Orin Industrial

NVIDIA also announced its latest Jetson AGX Orin Industrial module, which extends the capabilities of the previous-generation NVIDIA Jetson AGX Xavier Industrial and commercial Jetson AGX Orin modules. The Jerson AGV Orin Industrial module brings server-class performance to ruggedized systems. 

The latest module is meant to withstand extreme environments and extended shocks and vibrations. It can be used in a variety of applications, including farming, railway applications, mining, and even space applications. 

The module delivers up to 248 TOPS of AI performance with power configurable between 15-75 W. It supports multiple concurrent AI application pipelines with an NVIDIA Ampere architecture GPU, deep learning and vision accelerators, high-speed I/O, and fast memory bandwidth. 

Jetson AGX Orin Industrial also comes with an extended temperature range, operating lifetime, and shock and vibration specifications, as well as support for error correction code (ECC) memory.