[Music] the age of general purpose humanoid robots is officially here and things are moving fast Invidia just announced a wave of breakthroughs at their 2025 GTC conference including a powerful AI model called Groot N1 that's designed to make robots think plan and perform complex tasks with surprising Precision they've also teamed up with Google Deep Mind and Disney research to develop Newton a physics engine that's set to revolutionize how robots learn and interact with the world on top of that new humanoid robots from doot noetic Kepler and engine AI are pushing boundaries moving beyond basic tasks
to think plan and even tidy up your home so let's talk about it all right now during his keynote at the San Jose based event Nvidia CEO Jensen hang laid out a series of major breakthroughs that promised to supercharg humanoid robot development one of the headliners is the Isaac Groot N1 model touted as the world's first open fully customizable Foundation model for generalized humanoid reasoning and skills Wong was emphatic when he proclaimed the age of generalist robotics is here and showcased exactly why by demonstrating a humanoid robot from 1X performing domestic tidying tasks it deftly
picked up household objects and set them where they needed to go the really cool part was that all these movements were Guided by a post trained policy built built upon Gro N1 now Gro N1 is a GameChanger in humanoid Robotics and here's why it's built on a powerful dual system design that mimics how humans think and act on part called system 2 handles slower more deliberate thinking carefully planning actions and working through complex tasks the other part system one takes those plans and turns them into fast precise movements almost like a reflex Nvidia fed Groot
N1 an enormous amount of data combining real human demonstration data with massive amounts of synthetic motion trajectories generated in Omniverse this gave the model a remarkable ability to handle everyday tasks like grasping passing objects between arms or managing multi-step chores all without needing to be retrained from scratch Nvidia also announced a collaboration with Google Deep Mind and Disney research to develop Newton an open- Source physics engine built on the Nvidia Warp framework Newton is specifically tailored to help robots learn increasingly complicated tasks with improved realism and accuracy it's going to be compatible with Google deep
minds mujoo and Nvidia Isaac lab and there's talk of hooking it into Disney's physics engine as well this is especially intriguing because it could help developers replicate the intricacies of the physical world things like how objects bounce tilt or slide enabling robots to practice and refine their skills in Virtual environments before stepping into the real world see Disney research is one of the early adopters and they've got big plans for Newton they're already experimenting with Next Generation robotic character platforms such as their Star Wars inspired bdx droids which actually joined Jensen hang on stage at
GTC Kyle Laughlin senior VP at Walt Disney imagineering research and development explained that these bdx Droids are only the beginning The Mouse House wants to bring more animated characters to life in completely new ways making them more expressive and engaging than anything you'd see from a typical animatronic the combination of Disney's unique flare for storytelling plus the advanced Ai and physics simulation from Nvidia and Google Deep Mind suggests we're entering a new era of theme park experiences as exciting as these developments are the world of AI isn't just about robots it's also Al opening up
powerful opportunities for anyone willing to learn that's why we've partnered with growth school to bring you something valuable 2024 and early 2025 have been unpredictable jobs come and go and Financial Security is never guaranteed that's why multiple income streams aren't just smart they're essential AI can help and with the right skills you could earn an extra $10,000 a month now if you're wondering how to get started growth school has something really cool they're offering a 3our Hands-On AI training well you'll learn to use over 25 powerful AI tools normally it's paid but the first 1,000
AI Revolution viewers can join for free using the link in the description on top of that you'll get $500 worth of bonus resources just for signing up the training covers everything job hunting tips salary negotiation mastering Excel even content creation and it's not just for Tech experts whether you're in finance sales mark HR or even still studying this can work for you growth school has already helped millions of people level up and this could be your turn to stay ahead in an AI driven world so if this sounds like your kind of thing hit the
link below to grab your free spot plus don't miss joining growth schools WhatsApp Community it's a great place to connect with others diving into AI too all right now Nvidia has also taken the wraps off something called the Isaac Groot blueprint which is a system for generating synthe IC data for robot training to put it in perspective when robots learn from real human demonstration data alone there's only so much you can capture in say a single day but with the Groot blueprint for synthetic manipulation motion generation Nvidia was able to produce 780,000 synthetic trajectories in
just 11 hours that roughly translates to 6,500 hours about nine consecutive months of real human labor captured overnight and by mixing that that synthetic data with a smaller amount of real data the performance of the Groot N1 model improved by about 40% it's a mindboggling jump in efficiency and it points to a future where robotic systems can be trained to do complex tasks much faster than ever before in case you want to adapt the model for your own robot or environment Groot n1's data set is available for download on hugging face Nvidia basically wants developers
around the world to get their hands on this technology customize it and push the envelope another resource for that is the newly revealed Nvidia dgx spark personal AI supercomputer which provides a turnkey system for running all these Advanced AI workloads without hiring an entire data Center's worth of specialized infrastructure the Newton physics engine isn't out quite yet but the word is that it'll be available later this year meanwhile Nvidia Google Deep Mind Disney research and even a company called intrinsic are building open standards pipelines and best practices around open USD to make robotic data workflows
easier to manage the end goal is to reduce the friction for robotics developers everywhere letting them spin up simulations generate data train models and test them in real or synthetic scenarios without huge overhead on top of all that we're seeing new players from around the globe building their own humanoid robots spurred on by these Nvidia breakthroughs one example is the doot humanoid from China which features a whopping 28 degrees of freedom for Ultra precise dexterous manipulation doot claims it can pick up something as delicate as a cherry all while walking with a straight knee design
that saves up to 42% of its energy usage then there's the notic N2 humanoid robot which comes with about 40 tops Terra operations per second of AI computing power and a relatively modest 15 wat power consumption it's got 8 GB of video memory courtesy of an Nvidia Jetson module inside despite only having 18° of Freedom the N2 apparently can hit speeds of 3.2 m/s about 7.1 mph avoiding obstacles using deep reinforcement learning it stands just about 110 cm tall around 43 in and weighs around 20 kg about 44 lb giving it a lightweight form factor
perfect for weaving through tight spaces that might be too cramped for a larger machine or even a person another newcomer is the Kepler K2 humanoid robot standing at 178 cm tall about 70 in and weighing 85 kg about 187 lb with 52° of Freedom overall it's got 12 dedicated to ultra dextrous hands and the rest spread across its body for smooth human-like motion its nebula system handles Vision mapping multimodal interaction and precise prise hand eye coordination while also allowing the K2 to lift heavy objects and handle rough terrain Kepler is doubling down on the concept
of pairing a big cloud-based language model with a smaller faster specialized model running locally on the robot they've also built something called Kepler opsource a global developer platform that gives remote programmers the tools to code debug and manage multiple robots all at once finally we have engine AI pm01 robot which was shown dancing in a video that had some people questioning if it was CGI it stands at about 138 cm tall about 54 in weighs 40 kg about 88 lb and uses 23° of Freedom including five in each arm plus one in the neck to
bust out surprisingly fluid moves the waist can twist 320° and the joint torque can climb to 330 nton M about 243 ftlb letting it do Dynamic turns without tipping over it walks around 2 m/ second about 4.5 mph powered by a dual chip system with an Intel n97 and an Nvidia Jetson Orin the battery apparently clocks in at 10,000 milliamp hour lasting about 2 hours with typical usage engine AI says it's aiming for a price around $113,000 though whether or not that price sticks remains to be seen seeing how all these machines are emerging in
the same same year that Nvidia Google deepmind Disney and others are collaborating on open- Source Frameworks suggests we're entering a pivotal moment previously there might have been a handful of well-known robotics companies Boston Dynamics agility robotics maybe a few others showing off impressive demos but now the technology to simulate train and deploy humanoids is trickling down to more and more creators thanks to things like Gro N1 being open and fully customizable plus the Newton physics engine that's on its way the barrier to entry is dropping fast there's no doubt challenges remain real world tasks are
messy unpredictable and often demanding yet with nvidia's Groot blueprint generating massive amounts of training data Disney's animation expertise pushing robotic characters forward and startups building specialized humanoid robots progress is accelerating fast the Nvidia dgx spark personal AI supercomputer is lowering the entry barrier even further giving developers the tools to refine these models without costly setups for once the hype seems to match reality so what are your thoughts on nvidia's new breakthroughs and the rapid rise of humanoid robots are we closer than ever to seeing them in our daily lives or is there still a long
road ahead feel free to share your thoughts below and if you enjoyed this don't forget to like And subscribe for more thanks so much for watching and I'll catch you all in the next video
