Nvidia unveiled new robotics-related hardware and services during the festivities at its fall 2022 GTC conference that are geared toward businesses creating and testing machines in sectors like manufacturing. The robotics simulation platform Isaac Sim from Nvidia will soon be accessible on the cloud, the business announced. A new platform named IGX and the Jetson Orin Nano, a system built for low-powered robots, are both being added to Nvidia’s array of system-on-modules.
Designers may mimic robots interacting with mockups of the real world using Isaac Sim, which went into open beta last June (think digital re-creations of warehouses and factory floors). To train the models on actual robots, users can create datasets from simulated sensors, utilising synthetic data from batches of concurrent, distinctive simulations to enhance the model’s performance.
It’s not solely advertising hype, necessarily. According to certain study, companies trying to operationalize AI may be able to overcome many of the development issues by using synthetic data. Recently, MIT researchers discovered a way to categorise photographs using synthetic data, and almost every significant manufacturer of autonomous vehicles uses simulation data to enhance the real-world data they gather from moving vehicles.
Currently accessible on AWS RoboMaker, Nvidia NGC, from where it can be deployed to any public cloud, and soon on Nvidia’s Omniverse Cloud platform, Nvidia says the upcoming release of Isaac Sim will feature the firm’s real-time fleet task assignment and route-planning engine, Nvidia cuOpt, for improving robot path planning.
Teams may work remotely while sharing a virtual environment to simulate and train robots thanks to Isaac Sim in the cloud, according to a blog post by Nvidia senior product marketing manager Gerard Andrews. Developers won’t need a powerful workstation to do simulations if they use the cloud to run Isaac Sim. Simulated outcomes can be set up, managed, and reviewed on any device.
Nano-Jetson Orin
Nvidia first unveiled Jetson Orin, the newest iteration of its Arm-based single-board PCs for edge computing use cases, back in March. The Jetson AGX Orin was the first in the series, and Orin Nano broadens the selection with more reasonably priced options.
The aforementioned Orin Nano has the smallest Jetson form factor to date and can perform up to 40 trillion operations per second (TOPS), which is how many computations the chip can accomplish when it is fully utilised. It belongs to the Jetson family’s entry-level segment, which also contains six production modules built on the Orin platform and aimed for a variety of local, offline computing and robotics applications.
The Orin Nano supports AI application pipelines with Ampere architecture GPU, Nvidia’s 2020-launched GPU architecture, and comes in modules compatible with the company’s previously announced Orin NX. The Orin Nano 8GB, which offers up to 40 TOPS with power customizable from 7W to 15W, and the Orin Nano 4GB, which reaches up to 20 TOPS with power options ranging from 5W to 10W, will both be available in January starting at $199.
In a statement, Deepu Talla, VP of embedded and edge computing at Nvidia, stated that “over 1,000 customers and 150 partners have embraced Jetson AGX Orin since Nvidia announced its availability just six months ago, and Orin Nano will considerably expand this adoption.” (The Jetson AGX Orin costs well over a thousand dollars, which is obviously a significant difference from the Orin Nano;) Jetson Orin Nano “sets new standard for entry-level edge AI and robotics” with an orders-of-magnitude gain in performance for millions of edge AI and [robotics] developers.
IGX
Nvidia previewed IGX, a platform for “high-precision” edge AI, with a focus on industrial and logistics applications, in news that almost went unnoticed by us. The company claims that in highly regulated settings like factories, warehouses, clinics, and hospitals, it offers an extra layer of safety and low-latency AI performance.
An AI processor for autonomous industrial equipment and medical devices, the IGX Orin, is part of the IGX platform. According to Nvidia, developer kits with integrated GPU and CPU as well as a software stack with security and safety features that can be programmed and configured for various use cases will be made available early in the next year for businesses to prototype and test devices.
In order to provide IGX with full-stack, long-term support, Nvidia claims it is collaborating with operating system partners like Canonical, Red Hat, and SUSE.
According to Nvidia CEO Jensen Huang, “industries are adopting new functional safety criteria for AI and computing as people deal with robots more frequently.” The next generation of software-defined industrial and medical devices that can safely function in the same environment as people will be built by corporations with the aid of IGX.
