“In the realm of 3D printing and computational technology, the groundbreaking fusion of Artificial Intelligence (AI) and data is redefining our approach to problem-solving. It’s the modern alchemy,” explained Rev Lebaredian, Vice President of Omniverse and Simulation Technology at Nvidia, during the recent Hexagon Live 2023 event. Insight on how neural networks are being repurposed for practical application could not be more topical.
In a pivotal moment in 2012, Lebaredian recalls, a few grad students wielding consumer gaming PCs – or, as he refers to them, “personal supercomputers” – utilized a significant volume of internet data to unravel an age-old puzzle: differentiating cats from dogs in images. This simple feat might seem trivial, but it represented a monumental shift in the application of AI, more of this later.
Nvidia was one of many companies to join Paolo Guglielmini, Hexagon‘s dynamic CEO, who warmly greeted over 3,600 attendees hailing from 77 countries at Hexagon Live 2023 in Las Vegas. The guest list included representatives from Microsoft, AWS, Red Bull Racing, Stratasys, and 3D Printing Industry; in attendance to hear it all.
“We’re here to talk about your innovation, the phenomenal problems that you’re going to tackle using Hexagon technologies,” said Guglielmini, who emphasized his roots as an engineer and his appreciation for problem-solving. The CEO shared his experience working on the Large Hadron Collider (LHC) project at CERN. He referred to the LHC as “the world’s largest fridge” due to the extreme cooling required for the operation of its massive magnet systems, demonstrating the need for creative problem-solving in the field of engineering.
Guglielmini sees the same potential for innovation at Hexagon that he experienced at CERN, describing it as “optimism at scale.” He outlined Hexagon’s aim to digitize everything and create “digital twins” of everything from small components to entire cities. “We wanted to build digital twins of anything from small components, all the way to entire assemblies, all the way to industrial facilities and cities,” he said. These digital twins are not just virtual models but the foundation of an improvement journey towards more quality, productivity, and competitiveness.
Read more about Hexagon Live 2023 here.
Omniverse: an Industrial Metaverse
Hexagon Live 2023 was also the stage to reveal a major evolution in its already fruitful partnership with Nvidia. “Hexagon and Nvidia have been partners already for a very long time across our businesses. And you know how much it has been in the press and in the spotlight in the last couple of months because of advances in AI and the role that their technology play to accommodate for the explosion of artificial intelligence algorithms,” said Guglielmini.
According to the Hexagon President, the collaboration is now set to deepen, focusing on the realm of reality capture and the burgeoning industrial metaverse—a virtual reality space where users can interact in a physically plausible, shared environment. The goal is to enhance Hexagon’s capabilities in visualization and collaborative platforms, utilizing Nvidia’s flagship Omniverse platform. Omniverse is a real-time, photorealistic collaboration platform for 3D production pipelines.
“Don’t take my word for it,” quipped Guglielmini, inviting the CEO of Nvidia to elaborate on the groundbreaking collaboration.
Jensen Huang, co-founder, president, and CEO of Nvidia Corporation, reciprocated Guglielmini’s excitement. “Industries are racing to digitalize their physical processes. The opportunity is immense. The manufacturing industry alone is $46 trillion,” explained Huang.
According to him, digitalization is the magic bullet that can enable these industries to become software-defined, thereby unlocking the next frontier of automation: robotics. In the face of a future where everything that moves will be autonomous, Huang insists that training these robots requires learning in virtual worlds—a field where Omniverse has a unique selling proposition.
Huang explained, “Omniverse is a virtual world simulator for AI-enabled industrial applications that enables you to aggregate and simulate ultra complex data from specialized tools such as HxDR, and Nexus.”
This alliance with Omniverse, he contended, would bring reciprocal benefits. Hexagon, with its expertise in digital reality, geospatial reality capture, sensors, software, and autonomous technologies, would augment the Omniverse ecosystem. The ultimate beneficiary, he insisted, would be their customers, who would find it easier and faster to build, simulate, operate, and optimize these virtual worlds.
“And this is just the beginning,” said Huang, a hint towards a potential future suite of applications and services built on Omniverse. As Hexagon gears up to usher in a new era of AI-enabled digitalization, the road ahead promises to be one filled with technological breakthroughs and thrilling developments.
Burkhard Boeckem, CTO of Hexagon: Vision for the Industrial Metaverse
Continuing on the topic of the industrial Metaverse, Burkhard Boeckem, CTO of Hexagon, delineated the firm’s ambitious tech strategy, “Technology is evolving at an ever-increasing pace, and it’s sometimes hard to distinguish what is hype and what is real.”
Boeckem proclaimed, “We are building AI, creating an industrial Metaverse with advanced spatial computing.” The vision for the project, which is not expected to be fully realized for a number of years, “will be active in robotics and committed to sustainability.”
Hexagon’s AI capabilities extend beyond conceptualization. Boeckem shared, “We have incorporated AI in numerous applications. We use AI to segment digital reality data into land-covered layers, monitor change and progress for construction, model 3D environments and buildings, and even see what is underground.” A comprehensive blend of artificial intelligence and real-world applications may give Hexagon clients a competitive edge.
Hexagon and strategic partners such as Nvidia, AWS, and Microsoft have high hopes for the project. “The industrial Metaverse is unifying multiple technologies. It’s all about spatial computing, visualization AI, and digital reality technologies that converge to form a truly immersive experience. Our spatial computing technology is key to accelerating the industrial metaverse, and we can seamlessly connect the physical and digital worlds,” explained the CTO. “The time to innovate is right here, right now.”
Why build an Industrial Metaverse?
Insight into how to build the industrial Metaverse came from Rev Lebaredian, Vice President of Omniverse and Simulation Technology at Nvidia, who provided a glimpse into the future of virtual reality and the bold ambition of Nvidia to simulate the real world, not for entertainment, but for problem-solving and efficiency.
Lebaredian, who started his journey in visual effects for companies like Warner Brothers and Disney, joined Nvidia 21 years ago with a dream: to leverage computation and simulation to create real-time, immersive experiences. This path of inquiry eventually led to an unexpected revelation. “About 10 years ago, we discovered that the computing platform we built was perfect for this new kind of artificial intelligence. Deep Learning,” he explained.
The technology, which initially served to create fantastical virtual worlds for gaming and film, is now being repurposed for an audacious endeavor. “We can apply [this technology] to creating worlds for our AIs or artificial intelligence. The ones we want to create for them to learn it for us to test them inside these worlds before we bring them out into the real world” shared Lebaredian.
In essence, Nvidia’s vision is to transpose the real world into a virtual format to simulate and predict real-world outcomes, a process they call the Omniverse. This process, Lebaredian argues, unlocks “potential superpowers” by transforming everything into a software problem, applying the agility of software development to real-world challenges.
This isn’t about creating a perfect copy of our world; it’s an approximation but a close one. “We have algorithms; we have computing systems; we have things that problems that can be solved with approximations that are more than good enough for those purposes,” explained Lebaredian.
According to Lebaredian, the next challenge is two-fold: first, aggregating data from various sources to form a harmonious, cohesive virtual world; second, applying AI and simulation to this world for prediction and optimization.
This approach can potentially revolutionize industries with complex logistics, like car manufacturing. Factories that cost billions of dollars and employ thousands of engineers could significantly cut their operational inefficiencies and associated costs through this technology.
Lebaredian paints a picture of the process: “When they’re designing these factories, you have 1000s of people with different skill sets defining different aspects of the factory… Unfortunately, it’s impossible in the current way that these things are designed to ensure that it’s going to work the day and decide to build it because they’re all building these things or designing their parts in an island.”
The Omniverse could enable manufacturers to spot and solve problems before they physically materialize, potentially saving “billions of dollars” and reducing waste, contributing to a more sustainable future.
A Monumental Shift in the Application of AI
So, what about that seismic shift in AI’s application, telling the difference between cats and dogs?
“Previously, it was an obstacle that had long since stumped computer scientists. It was an area many of us had somewhat given up on, assuming that such a task would not be possible within our lifetime,” said Lebaredian. But with this breakthrough, he continued, “We realized that the key thing that had changed was these algorithms that are developed now; the most advanced ones aren’t going to be created by humans directly.”
In essence, Nvidia had started writing software to write software. Feeding large volumes of data, their algorithm produced a new algorithm. The next logical step, according to Lebaredian, was to find an abundant source of “life experience” or data to feed into the model.
Simulated virtual worlds, already being developed for video games, presented an ideal resource. “We’ve already been simulating virtual worlds for entertainment purposes for video games. Well, let’s adapt those things to AI,” he suggested.
The fruits of this adaption are astounding, resulting in a monumental change in how software is created. “Now we have computer vision algorithms that can tell the exact species of cats and dogs far better than any human, even the best connoisseurs at this particular skill,” Lebaredian remarked.
He went on to describe Nvidia’s partnership with Hexagon. In a recent collaboration, the two tech giants developed an AI-enabled web application based on Omniverse, Nvidia’s platform. It offers users an accurate comparison between their digital twin and physical twin, effectively accelerating decision-making and optimizing planning and operations.
Lebaredian added, “What we’re designing with Omniverse is a system that can leverage and take advantage of all of the computing power available in the data center.” The software scales to the problem, a concept that completely upends our current understanding of computational capacity.
AI’s ability to create virtual worlds could unlock unprecedented possibilities
Lebaredian remarked, “Earlier I mentioned how we started the journey building Omniverse so that we could create worlds to train and test our AIs…But we also believe that building virtual worlds themselves or building worlds is extremely difficult.” Addressing this challenge, Nvidia proposes to harness AI to augment our capacity to create complex virtual environments.
In today’s world, the skill to construct sophisticated virtual landscapes like those seen in popular video games such as Fortnite or Roblox is a rare one, nurtured over decades. However, the ambition to create digital twins of the real world to improve it demands a dramatic expansion of this talent pool.
Lebaredian states, “We need to enable more people to be able to build such things. They can spend three decades becoming experts in virtual world-building. And to do that, we need AI. We need AI to assist us.” Therefore, a symbiotic cycle is envisioned: AIs trained within these digital ecosystems learn to perceive and manipulate these worlds, assisting humans in creating further virtual environments.
When probed about the novelty of this concept, Lebaredian agrees, “It is something fundamentally new to mankind. We haven’t built systems that were intelligent enough to improve themselves before.” This profound statement underscores Nvidia’s faith in the untapped potential of AI and its possible implications for the tech industry and society at large.
His final, modest comment, “We believe so; that’s why we started investing in over a decade ago. And it turned out to be a pretty good bet so far. We’ll see where it goes,” leaves us on the precipice of an AI-orchestrated future, full of fascinating unknowns.
What does the future of 3D printing for the next ten years hold?
What engineering challenges will need to be tackled in the additive manufacturing sector in the coming decade?
While you’re here, why not subscribe to our Youtube channel? Featuring discussion, debriefs, video shorts, and webinar replays.
Are you looking for a job in the additive manufacturing industry? Visit 3D Printing Jobs for a selection of roles in the industry.
Featured image shows Hexagon CEO Paolo Guglielmini. Photo by Michael Petch.