The recent release of footage showcasing a humanoid robot powered entirely by a single neural network has shocked the entire robotics industry. This groundbreaking development comes from 1X, a robotics company backed by OpenAI. The demo of their NEO robot is absolutely mind-blowing, as it uses a single neural network to control every behavior from pixels to actions, with no human intervention or scripted code.
1X’s mission is to provide safe, intelligent androids for physical labor, designed in the human form for maximum generality. The company has raised a significant amount of funding, indicating a strong support for their vision of bringing AGI to robotics.
The video demo showcases the incredible capabilities of the NEO robot, controlled by a single vision-based neural network that emits actions at 10 Herz. The robot’s movements are autonomous, seamless, and controlled purely by AI, with no cuts, CGI, or scripted trajectory playback.
This development is indicative of the rapid advancements in the robotics industry, with companies like 1X, Tesla’s Optimus, and others pushing the boundaries of what’s possible with AI and robots. The future of humanoid robots in everyday environments is becoming increasingly realistic, with the potential to assist with everyday chores. While some may have concerns about this future, the advancements in robotics and AI are truly exciting.
The video also highlights other remarkable robots, such as a bipedal robot from a PhD student at UC Berkeley, an open-source collision avoidance robot, and a dog-form factor robot from a company called Fox Glove.
Overall, the advancements in robotics and AI are paving the way for a future where humanoid robots could become an integral part of everyday life. The possibilities are endless, and the pace at which technology is evolving is truly remarkable. This video serves as a testament to the incredible potential of AI-backed robots.
Watch the video by Matthew Berman
Video Transcript
1X a robotics company backed by open AI just dropped an incredible demo of their humanoid robot it uses a single neural network to control the robot from start to finish and it is absolutely mindblowing if you remember my video about mobile Aloha this video takes that idea even further check out this demo 1X promises to bring AGI to Robotics and this is Eric Jen from 1X here’s our latest software update on 1X AI every Behavior you see in this video is controlled from pixels to actions with a single neural net architecture no T op which is T operation which means no
Human is controlling it no scripted replay or task specific code no CGI all-in-one continuous video shoot and as mentioned 1X is backed by open Ai and just about a month ago they raised $100 million to build humanoid robots so here’s the blog post about how they did
It let’s take a look let’s talk about their mission first onex’s mission is to provide an abundant supply of physical labor via safe intelligent Androids our environments are designed for humans so so we design our Hardware to take after the human form for maximum generality now between this and Optimus from Tesla
Humanoid robots are coming quickly and when you mix Ai and robots together you get magical outcomes every Behavior you see in the video is controlled by a single vision-based neural network that emits actions at 10 Herz the neural network consumes images and emits actions to control the driving the arms
Gripper torso and head the video contains no tell operation no comp computer Graphics no cuts no video speedups no scripted trajectory playback it’s all controlled via neural networks all autonomous all 1X speed and to train the models they assembled a highquality diverse data set of demonstrations
Across 30 EV robots and then they use that to train a base model that understands a broad set of physical behaviors from cleaning to tidying homes to picking up objects to interacting socially with humans and other robots we then fine-tune the model into a more specific family of capabilities and then
Fine-tune those those models further to align the behavior with solving specific tasks this strategy allows us to onboard new skills in just a few minutes of data collection and training on a desktop GPU and again if you remember back to my mobile Aloha video you can build this yourself technically mobile Aloha is
Fully open source the hardware the software the training set everything I’ll link that in the description below and it seems like there are a ton of companies working on robotics right now check out this incredible bipedal robot from this PhD student at UC Berkeley N And here’s another example of a robot with an advanced open-source Collision detection algorithm look how incredible this [Applause] Is [Applause] [Applause] [Applause] and this is the paper behind it agile but safe learning Collision free high-speed legged Locomotion so you can read the paper yourself I’ll drop a link in the description below and last look at this dog form factor robot from a company called Fox Glove [Applause] [Applause] Dev what kind of trick does it do hello hello [Applause] wow [Applause] and you can really see how robots are advancing quickly both in indoor environments and outdoor ruggan environments Boston Dynamics really paved the way for all of these new robotics companies but now that large language models are proliferating and
Evolving at the speed that they are so many more robotics companies are launching with absolutely mind-blowing demos I’m still waiting for the day that I can have a humanoid robot in my house helping me with everyday chores is that something you want also let me know in
The comments below I know a lot of people can be pretty afraid with that kind of future but I’m personally very excited if you liked this video please consider giving a like And subscribe and I’ll see you in the next one
Video “OpenAI-backed “AGI ROBOT” SHOCKED The ENTIRE Industry” was uploaded on 02/09/2024 to Youtube Channel Matthew Berman
The post “Industry in Awe as OpenAI-Backed “AGI Robot” Stuns Everyone – Video ” by GretAi was published on 02/17/2024 by news.gretai.com