GTC November 2021 Keynote with NVIDIA CEO Jensen Huang
NVIDIA is already building and offering a metaverse environment for a variety of uses, embodied in its NVIDIA Omniverse platform. NVIDIA Omniverse is a platform that includes 3D animation tools, which the company calls the "metaverse" for engineers. It can be used for a variety of purposes, including making 3D copies.
The Omniverse Enterprise platform for 3D collaboration across organizations has been a special focus for NVIDIA since its announcement in April 2021. It allows interaction in the metaverse using physically accurate digital twins, such as products under development or problems being solved. It can be a car, a factory, or a retailer's warehouse, or even an entire city - the platform allows you to create a “digital twin” of anything. Even the Earth.
In the future, the data will be collected in the Earth's digital twin, created in the Omniverse, and a set of physics models with machine learning will allow predicting climate change.
Also noteworthy was the announcement of cooperation with Lockheed Martin - the companies plan to create the world's first laboratory based on AI and Omniverse, which will be engaged in modeling and forecasting forest fires.
“__A theme you will see constantly is how Omniverse is being used to create digital twins of warehouses, factories and factories, physical and biological systems, 5G peripherals, robots, self-driving cars and even avatars” - noted Jensen Huang.__"**
Since the launch of the beta version of the platform to date, more than 70,000 individual users have downloaded it and are used by more than 700 companies, including BMW and Ericsson, according to NVIDIA.
NVIDIA Omniverse Avatar - has become one of the key innovations presented at the GTC. The solution integrates NVIDIA technologies in speech AI, computer vision, natural language understanding, recommender systems, and simulation technologies. The avatars created on this platform are interactive characters with 3D graphics based on ray tracing.
The Omniverse Avatar allows you to create smart assistants that are easy to customise for almost any industry, such as interacting with customers when ordering in a restaurant, banking, making in-person appointments, making reservations, and more.
The company also announced NVIDIA Omniverse Replicator, a powerful synthetic data generation engine that generates physically correct synthetic data for training deep neural networks. It is part of the Omniverse platform.
It also allows the generation of data that is difficult or impossible for humans to mark: for example, speed, depth, closed objects, weather inclemencies, or tracking the movement of objects between sensors.
An interesting update appeared also in the software NVIDIA Riva for the creation of interactive services of AI - Custom Voice. This new feature allows companies to develop with Riva voice for enterprise services in hours instead of weeks and using small amounts of data.
"Riva delivers world-class speech recognition accuracy and its response time unrivaled. In half an hour of training, she can speak in a specific voice: for example, the voice of one of your brand representatives" - says Jensen Huang.
Also introduced was the NVIDIA Modulus framework for creating and training physics models that obey physical laws. This tool is primarily intended for engineers, scientists and researchers. It can be used, for example, in the field of protein engineering or climatology.
To help deliver NVIDIA AI-powered services, Jensen Huang announced the expansion of the NVIDIA Launchpad program, which offers organizations an end-to-end platform for AI tasks.
Another important announcement is the release of the Jetson AGX Orin, which the company calls the world's smallest, most powerful and energy efficient AI supercomputer. It is designed for robots, autonomous machines, medical devices, and more. The Jetson AGX Orin is a sequel to the Jetson AGX Xavier, which the company released in 2018. Compared to it, the Jetson AGX Orin has six times the processing power in the same form factor. The computer can fit in the palm of your hand. Also, the new product is compatible with its predecessor computer.
The manufacturer claims that the Jetson AGX Orin __is capable of performing 200 trillion operations per second. It is powered by NVIDIA Ampere GPUs, ARM Cortex-A78AE processors and next-generation deep learning and computer vision accelerators. This makes it possible to run many AI applications.
NVIDIA has unveiled the cuQuantum development kit to accelerate the simulation of quantum computers on classical systems. Such modeling helps scientists to quickly create and test new quantum algorithms that modern hardware solutions will not "pull". Simulation is also critical for validating and testing the performance of the next generation of quantum hardware solutions.
Unmanned vehicles
Transportation technology remains one of the hottest topics, and NVIDIA is not bypassing it at its events. In addition to the aforementioned Omniverse Replicator-powered NVIDIA DRIVE Sim app, the GTC showcases NVIDIA DRIVE Concierge and DRIVE Chauffeur, two AI-powered platforms designed to eliminate the stress of daily driving.
The software-defined vehicles will include two key components built using the NVIDIA DRIVE Orin SoC: one for AI assistants, DRIVE Concierge, and one for autonomous driving, DRIVE Chauffeur. DRIVE Concierge uses the NVIDIA DRIVE IX platform for self-driving cars, as well as the NVIDIA Omniverse Avatar, and works directly with DRIVE Chauffeur.
DRIVE Concierge supports features such as automatic parking, digital assist, reliable situational awareness, and more. DRIVE Chauffeur supports autonomous driving on different roads - in the city and on highways, relieving the driver's burden of controlling the vehicle and monitoring the situation on the road.
Health care
The medical industry is one of the industries adopting AI and high performance computing for mission-critical tasks. For this industry, NVIDIA also announced new products at the GTC conference. One is NVIDIA Clara Holoscan, a new computing platform based on the NVIDIA AGX Orin mini-supercomputer, which provides a scalable software-defined computing infrastructure for processing data streams from medical devices. For example, with an endoscope or computed tomography equipment.
«The company's policy and philosophy is to create only what does not exist yet. Why would we support such a large resource of 22 thousand professionals in order to repeat already created developments? - noted the CEO of NVIDIA. - Everything we do turns out to be a full stack of products, and what we do is open source. This is why we have been able to cover so many industries and clients. Our technology is in every cloud and every computer manufacturer. We are partners with them. NVIDIA helps them with system architecture, integration, software, and we're even opening up new markets for our OEM partners so we can grow together.»__* - Jenshen Huang
