Text/VR Gyro Small Drill Wind
"The NVIDIA Omniverse platform, which is suitable for virtual worlds, digital twins and robotic systems, will set off the next wave of AI." By this gth, Jen-Hsun Huang's vision of Omniverse was in full swing, pointing to the next wave of AI.
Compared with gtccock a year ago, Omniverse's display case is not much different, but with omniverse's update and improvement, its operation and scene details make the described future appear clearer, omniverse will play an important role in the next wave of AI.
Omniverse will set off a new wave of AI
NVIDIA's GTC opening is still full of AI cases, unmanned driving, agriculture, robotics, digital people, medicine, meteorology... AI seems to be penetrating across fields. Not only the demonstration of the case, Huang Jenxun also showed a number of NVIDIA's achievements in the direction of AI, including:
Announced language AI Riva
Full release of the AI model toolkit Maxine
The AI framework Merlin for the recommendation system
Sionna, an AI framework for 6G communication research
Nemo Megatron, an AI framework for training large language models
Modulus, an AI framework for developing Physics-ML models
......
Source: Nvidia
This shows the applicability of NVIDIA AI technology in many fields. In the follow-up speech, in addition to the release of the H100 engine and the relevant updates of Omniverse, the real-time virtual human Avatar, robot factory, and unmanned driving cases with full AI content directly opened up our future imagination.
Under the Omniverse Avatar architecture, a number of applications have been born, such as the "mini version of Huang Jenxun" Toy Jesen and Tokkio. Toy Jesen and Tokkio are smarter this year than last year's debut.
Toy Jesen appeared to tell how he was driven by NVIDIA AI layer by layer, and the Omniverse Avatar architecture was generated step by step, and became the appearance that can now be communicated with real humans.
Not only is it a matter of communicating freely, but in the follow-up presentation of Omniverse Cloud, Toy Jesen is involved in the design work of human designers.
Toy Jesen, who spoke to Wong In-hoon
Tokkio in the form of a semi-broken shell also appeared in GTC again this year, and Tokkio as a self-service AI assistant was not only introduced into retail stores, fast food restaurants and other scenarios, but also appeared in autonomous vehicles, playing a high-end version of Siri voice robot assistant.
Tokkio is not only able to communicate directly with users through voice, but also can complete many aspects of work from driving to life assistants, such as displaying the road conditions from the perspective of the car, automatically stopping, identifying specific objects, booking tickets, and so on.
In addition, what caught the attention of Gyro Jun and got the key demonstration of this gtcach is the robot, Huang Jenxun mentioned in his speech:
"The first wave of AI learns perception and reasoning, such as image recognition, speech understanding, recommendation videos or commodities, and the next wave of AI is headed by robots, that is, using AI to plan actions."
"Digital robots, avatars and physical robots will complete the perception, planning and action. Omniverse will also become an indispensable tool when making robotic software, and Omniverse will set off a new wave of AI. ”
"NVIDIA is building multiple robotics platforms — DRIVE for self-driving cars, Isaac for manipulating and controlling systems, Metropolis for autonomous infrastructure, and Holoscan for robotic medical devices."
"The demand for robots and automation is growing exponentially, and NVIDIA works with thousands of customers and developers to build robots that are widely used in manufacturing, retail, healthcare, agriculture, construction, airports and entire cities," Huang said. ”
He mentioned that one of the fastest growing areas of robotics is autonomous mobile robots (AMRs). According to it, autonomous mobile robots are essentially driverless cars indoors, with low speeds but highly unstructured environments, currently with tens of millions of factories, shops and restaurants, as well as hundreds of millions of square feet of warehousing and operations centers.
These large-scale factories of the future will be the domain of robots that learn and acquire skills in digital twins built by Omniverse, and then take to "jobs" to work with humans.
In his speech, Huang Renxun showed Amazon and PepsiCo factories as examples, he mentioned that digital twins are an indispensable part of the warehouse and factory of the future, and manufacturers such as Amazon are using Omniverse digital twins to reshape warehouse logistics end-to-end and greatly improve operational efficiency.
At the same time, Omniverse helped it better train autonomous robot sorting and picking solutions, training the perceptual systems of these robots to be accurate enough to avoid system failures.
In the above process, it can be seen that the digital twin, as a virtual world connected to the physical world, is extremely closely related to Omnires and AI.
Shi Chengqiu, senior technical marketing manager of NVIDIA China, said at the communication meeting, "Many parts of the virtual world are completed by AI, and the development of the virtual world is inseparable from artificial intelligence, and the development of digital twins is inseparable from artificial intelligence." ”
AI eventually needs a powerful processor to do all this acceleration, and this powerful processor is a series of stacks of Nvidia hardware and software, and in the end, Omniverse becomes this important bearer.
Shi Chengqiu also mentioned why NVIDIA now has the courage to do Omniverse that AI is an important driving force.
"The main reason is NVIDIA's technological breakthrough after 2018, on the one hand, the universalization of AI, which has reached a new level under the powerful computing power of GPUs; another important aspect is the major breakthrough in high-performance visualization operations, including Ray Tracing."
The advent of the AI wave and the robot era requires a large number of digital twins, and we can see in this GTC that the focus of Omniverse updates and releases is precisely on digital twins.
An important omniverse update is released
Drop digital twins
At this GTC, NVIDIA made 3 important releases closely related to Omniverse, and made 8 updates to Omniverse's existing applications and functions.
Three important releases:
The NVIDIA OVX computing system that drives the massive digital twin
Digital twin platform for scientific computing
Cloud service Omniverse Cloud
1. OVX calculation system
NVIDIA's OVX computing systems are designed to run complex digital twin simulations in Omniverse.
According to reports, the OVX system combines high-performance GPU-accelerated computing, graphics processing and AI with high-speed storage access, low-latency networking, precise timing, and the performance needed to create a realistic digital twin. OVX will be used to simulate complex digital twins to model buildings, factories, cities, and entire worlds.
2. Digital twin platform
Nvidia, an accelerated digital twin platform for scientific computing, consists of the NVIDIA Modulus AI framework for developing neural network models for machine learning in physics and the Omniverse 3D World Simulation Platform.
The platform enables interactive AI simulations based on physical information to accurately reflect the real world in real time, enabling simulations such as computational fluid dynamics to be 10,000 times faster than traditional engineering simulation and design optimization workflow methods. Compared to previous AI models, researchers are able to model complex systems, such as extreme weather events, with greater speed and precision.
3、Omniverse Cloud
Omniverse Cloud cloud services give artists, creators, designers, and developers instant access to the Omniverse platform, allowing users from billions of devices to collaborate and simulate 3D designs.
Omniverse Cloud includes three services: Nucleus Cloud, a simple "one-click collaborative" sharing tool that allows artists to access and edit large 3D scenes from any location without having to transfer large data sets; Omniverse Create, an app that enables technology designers, artists, and creators to build 3D worlds together in real time; and View, which allows non-technical users to view Omniverse scenes.
Compared to Omniverse Cloud, which is more used for collaboration, the OVX computing system and digital twin platform are undoubtedly aimed at digital twins. In addition to these three releases, this year's GTC also brings a series of updates from Omniverse.
For example:
Omniverse Kit and Create 2022.1 can build more advanced production tools, with a series of rendering and material improvements, and its advanced animation and behavior tools will make it easier for designers and others to connect virtual and reality in their scenes;
Omniverse View has built an interactive viewer to facilitate non-professional viewing and review, enabling it to comment and mark up, review and approve during the presentation process;
Nucleus (including Work Station and Work Server) has made major updates to user permissions, so that enterprise IT customers can have more access and control, so that more accurate resource allocation and adjustment based on the background;
Repliacator has become an extension and use of Omnipree, which can help designers quickly recall different materials, modeling, and digital assets in the material library in the digital twin virtual world, and can call and quickly generate the scenes they want;
Omniverse XR, on the other hand, will launch a new beta version that allows you to bring your 3D work to VR for viewing, and annotating, labeling, and manipulating scenes, objects, etc. in Omniverse. Vr and AR modes enable users to view already supported Omniverse digital assets and scenarios in a more immersive way.
Shi Chengqiu revealed that these eight updates have pushed the entire Omniverse forward, and in the one year between the two GTCs, the cross-industry applications connected by Omniverse's Connector have increased from 8 to 80.
As an important component of Omniverse, Connector can connect a range of third-party ISV (Independent Software Developer) tools and digital assets used on the Omnivierse platform and can be used across domains, whether it is digital twins, industrial automation, sensor models, robot platform integration can help its customers build industrial-grade digital twins.
In his speech, many developers want OEMs to connect Omniverse directly to their software suite, while Jen-Hsun Huang said he would release format converters for 26 manufacturing industries to connect industrial workflows to Omniverse.
As can be seen from the above, in terms of updates, Omniverse is iterating on functions and applications, while constantly accessing workflows in more fields; on the other hand, it is constantly consolidating digital twins. The update of Omniverse, the increasing maturity of digital twin technology, and the multi-field penetration of AI show that NVIDIA is fully prepared for the next wave of AI and digitalization.
Under the wave of digitalization
It's not just NVIDIA that beats the drums of war
"The NVIDIA Omniverse platform, suitable for virtual worlds, digital twins and robotic systems, will set off the next wave of AI. Just as TensorFlow and PyTorch are the basic frameworks for perception-oriented AI, Omniverse will also be an integral part of action-oriented AI. ”
AI, digital twins, Omniverse, from the underlying structure to business planning, in all aspects, NVIDIA looks like it is fully prepared for the era of AI robots, and Huang Jenxun has also set its sights on 6G.
"The 6G standard will come out around 2026, 6G is the trend of the times - thousands of robots and robots will become the middle army among network users, 6G is gradually forming a prototype around some basic technologies, like the network, 6G will be largely defined by software." Networks will be driven by AI, and a digital twin based on ray tracing and AI will help optimize networks. ”
The advent of 5G and even 6G has made the era of AI robots, which was once a distant place, shine into reality under the demonstration of Huang Jenxun, and industrial production will also usher in a wave of acceleration with the help of AI to achieve a larger area of digitalization and intelligence.
Not only NVIDIA, the pursuit of digitalization and intelligence has long attracted attention in China. In the 14th Five-Year Plan released this year, the term digitalization has been repeatedly mentioned, and efforts will be made in many directions to "accelerate digital development and build a digital China".
Image source: Network
Tencent's 2021 financial report also shows that it has listed digitalization as an important development direction. In Tencent's new brand commitment of "helping the real economy" in 2021, it can be seen that in 2021, Tencent will help promote the digital transformation of cities in various directions such as water conservancy, rail transit, and construction.
In 2021, "helping the real economy" has become Tencent's new brand commitment. We explore 60 meters underground to help the "Century-old Water Conservancy Project" revitalize the digital new machine; go deep into the rail transit to help build the "Greater Bay Area on the Track"; go to the field and let the whole life cycle of Dingxi potatoes bathe in digital sunshine; enter the factory assembly line, use AI to solve the dilemma of industrial quality inspection, and promote "Made in China 2025"; down to the construction site, explore the new digital model of "urban infrastructure"...
Next, Tencent will continue to explore cutting-edge "hard technologies", such as chips, robots, digital Homo sapiens, etc., and AI, robots, automatic driving and other multi-directional actions do not mention.
Looking at the national development strategy and the dynamic adjustment direction of enterprises in recent years, it can be seen that digitalization and intelligence have become a trend. The wave of digitalization is rolling in, and in the context of the consensus between technology and the market, perhaps the era of robots is in the near future.