NVIDIA And Alphabet Show Off Next-Gen Robot Tech Getting A Grip With AI

Autonomous robot picking up pieces of metal.
It may seem hard to believe, but here we are well into 2024 and robotic applications are still having trouble grasping objects with the same dexterity as humans. Weren't we supposed to have flying cars and mastered time travel by now? While some of the ideas concocted in Hollywood haven't exactly panned out (yet), NVIDIA, a master in artificial intelligence hardware and software, is helping robots get a grip on automation.

Announced at the Automate trade show in Chicago, Alphabet's software and AI robotics company, Intrinsic, has integrated NVIDIA AI and Issaac platform technologies (which NVIDIA unveiled at GTC 2024) to level up autonomous robotic manipulation. As such, NVIDIA and Instrinsic shared a couple of demos that appear rather simple on the surface, but are actually quite complex to achieve.

The demo above simulates a robotic arm picking up pieces of sheet metal from a bin and plopping them into and adjacent bin.

"NVIDIA Isaac Sim is an extensible robotics simulation platform that gives you a faster, better way to design, test, and train AI-based robots. It’s powered by Omniverse to deliver scalable, photorealistic, and physically accurate virtual environments for building high-fidelity simulations," NVIDIA explains.

Though it looks simple, NVIDIA says grasping is a sought-after skill in the field of robotics. Up to this point, achieving seamless grasping skills has been met with several challenges, including the cost to program solutions and difficulties in scaling. It's also incredibly time consuming.

The prototype you see in the above simulation leverages Intrinsic's Flowstate developer environment for AI-based robotics solutions. It includes a host of technologies, including NVIDIA's Isaac Manipulator, which is a collection of foundation models and modular GPU-accelerated libraries, and NVIDIA's Omniverse platform.

A second demo shows how NVIDIA's foundation model used in Intrinsic Flowstate can be adapted to an actual robotic arm. It's essentially the same demo—picking up metal pieces from one bin and placing them in another—but in a real-world application, not just a simulation.

"For the broader industry, our work with NVIDIA shows how foundation models can have a profound impact, including making today's processing challenges easier to manage at scale, creating previously infeasible applications, reducing development costs, and increasing flexibility for end users," said Wendy Tan White, CEO at Intrinsic, in a blog post announcing the collaboration with NVIDIA.

Simulation is key to achieving these goals, as it allows companies to design prototypes for customers before committing to the costs of actual hardware. It's a cost-saving step before actual deployment, and it's made possible through NVIDIA's expanding AI ecosystem.

As part of the partnership between NVIDIA and Intrinsic, the two will collaborate on state-of-the-art dexterity and modular AI solutions for robotic arms. This will entail a large collection of foundation models and GPU-accelerated libraries to fast track next-gen robotics and automation applications.