OpenAI's "FULLY AUTONOMOUS" Robot Just SURPRISED The ENTIRE INDUSTRY!
21:57

OpenAI's "FULLY AUTONOMOUS" Robot Just SURPRISED The ENTIRE INDUSTRY!

TheAIGRID 09.02.2024 29 053 просмотров 782 лайков

Machine-readable: Markdown · JSON API · Site index

Поделиться Telegram VK Бот
Транскрипт Скачать .md
Анализ с AI
Описание видео
✉️ Join Our Weekly Newsletter - https://mailchi.mp/6cff54ad7e2e/theaigrid 🐤 Follow us on Twitter https://twitter.com/TheAiGrid 🌐 Checkout Our website - https://theaigrid.com/ https://twitter.com/ericjang11/status/1755655698120405461 Welcome to our channel where we bring you the latest breakthroughs in AI. From deep learning to robotics, we cover it all. Our videos offer valuable insights and perspectives that will expand your knowledge and understanding of this rapidly evolving field. Be sure to subscribe and stay updated on our latest videos. Was there anything we missed? (For Business Enquiries) contact@theaigrid.com #LLM #Largelanguagemodel #chatgpt #AI #ArtificialIntelligence #MachineLearning #DeepLearning #NeuralNetworks #Robotics #DataScience

Оглавление (5 сегментов)

Segment 1 (00:00 - 05:00)

so what you're looking at is a fully autonomous robot that was just released in a video produced by a company called 1X robotics the open ey back to robotics company has produced the first videos in which we do see humanoid robotics performing a range of different tasks that are fully autonomous now many people truly and I mean truly cannot understand the ramifications for this video because they are staggering we are witnessing the first kinds of developments that will lead to the creation of physical AGI and it's right here for you guys on this channel now I'm really glad that I'm able to document and see this because this process to me is fascinating we're witnessing robots doing things using endtoend neuron Nets fully autonomously now one of the things that I did see that many people didn't realize was the fact and is the fact that this is one time speed now essentially why that is so crazy is because many of you who've seen robot demos before have known that one time speed is not something that they usually do and this is because the autonomous modes in robots is usually very slow for example in Google's robot demo they had this robot and in the top left hand side you can see that it says four times speed and of course it does this because of course right now it is very hard to get robots to move quickly in real time and that why this demo from 1X robotics is so impressive I mean if we do take a look at the range of tasks being done here there is so much to uncover including some very new things from their blog posts that we are going to dive into so what of the things I wanted to dive into was one of the key actions that these robots were actually doing and if you didn't pay attention I'm going to rewind the video so you know and that's for this section right here we can see that these robotics the robots are actually called EES that that's the make of this robot and essentially they're plugging themselves in and I'm assuming that they are charging themselves which is a very good thing to be able to do autonomously because if you have a robot you don't want to have to be able to charge it all the time you're going to want it to be able to autonomously be able to detect when its battery levels low and then of course plug itself in and continue doing whatever tasks it may do now that doesn't of course spell the end or the start of the Terminator era but it does show us that things are really rapidly improving now many other robotics companies have shown us their demos but to my surprise this is one of the first companies that is really doing some autonomous stuff and based on their history this is also no surprise too now one of the things that is really cool about these robots is that their hands are much different than some of the other humanoid robots that we've seen they're using a different approach that was quite similar to what we saw in the mobile Aloha demo so essentially when these robots are gripping things for examples these doors and of course plugging themselves in we're seeing them do it with these grippers okay um and I find that this is quite fascinating as well because it shows us that there's a lot you can do with things that aren't human hands what was also quite shocking as well was just the Staggering amount of Eve robots that these guys do have on site which are clearly ready to be deployed but I'm sure that they're probably going to be undergoing even more training to be general purpose to cover a wide range of tasks so if we do take a look at the video I find additionally as well the first section of the clip where we're seeing these robots pack things to be a very bold statement because it shows us where some of these humanoid robots are going to be in the future and that is of course no surprise because if you've seen some of the demo videos of this robot you'll know that these robots are essentially designed for that use purpose now essentially what they do talk about in their blog post is the fact that this is based on a Vision based neural network so it says every Behavior you see in the previous video is controlled by a single Vision based neural network that emits actions at 10 Hertz per second so essentially this indicates that a singular AI system specifically a neural network that relies on visual data likely from cameras or sensors that simulate eyes is responsible for deciding and guiding all of these actions and neuron networks are a type of AI that mimic the way human brains operate allowing the robot to learn and adapt from its surroundings and being Vision based means that it uses these visual inputs to make decisions now the 10 HZ part means that the neuron Network updates the robots actions 10 times per second and Hertz is a unit of frequency and in this context it signifies how frequently the neuron Network sends commands to the robot and emitting actions at 10 htz allows the robot to respond to its environment relatively quickly making adjustments to Its Behavior based on the neural network calculations at a rate of 10 times

Segment 2 (05:00 - 10:00)

every second now later on in the video I'm going to be showing you guys some other clips that they did tease but for right now I do want to talk about one thing that did peque my interest which was software Engineers 2. 0 so they said all of the capabilities shown in the video were trained by our Android operators they represent a new generation of software 2. 0 Engineers who Express robot capabilities through data instead of writing code and essentially what that means is that this introduces the concept of software 2. 0 and it with software 1. 0 essentially where Engineers write lines of explicit code to tell a robot exactly what to do in every possible situation for example if you see a red ball move towards it blue cube turn left Etc and so on now essentially in software 2. 0 the robot learns by example and operators provide lots of data like images or videos demonstrating the desired skill and the robot's neuron Network analyzes this data to find patterns and learn how to do the skill on its own now I find that to be really fascinating because this is going to be a growing sector now in addition there were many different comments about this that I found rather interesting also and this is Dr Jim fan a senior AI research scientist at Nvidia Ai and he said that the first skill that any robot should learn is to plug and unplug themselves from power roombot is able to do it and now humanoids 2 one fewer task that the human babysitters need to worry about congrats to Eric Jang and the 1A X team now if you don't know who Eric Jang is this is of course the guy who is the VP of AI at 1X Technologies now essentially he said every Behavior you see in this video is controlled from pixels to actions with a single neuronet architecture no teleop no scripted replay or task specific code no CGI all inone continuous video shot now if you don't know why he's stating those things it's because number one no teleop essentially means that there's no teleoperated controller meaning that this video the one you're seeing here where there doing all of these crazy actions means that it's the robot doing it by itself and there isn't a human operator behind it we actually did see a lot of teleop last time when we looked at things like mobile Aloha and of course whilst there were autonomous cases many people were commenting on the fact that some of it was teleoperated in addition he stated that it is not scripted replay or task specific code and it's not CGI because of course there are some companies out there that do have Demos in which it is a scripted replay which essentially means that it's all set up just for the video or there is Task specific code which essentially means that it's just for that task this is a neuronet that knows exactly what to do now I did find one of these comments rather hilarious but I did want to add it into the video because Eric Jang did actually make a comment on it too so he said this is humans in costumes and of course someone responded to that statement saying they look so real but trust me there's no humans inside and I think this is the first inkling which shows us that in the future some of these humanoid robots might be so realistic that some people can't tell and don't believe that it is completely AI that is doing the work now essentially what was also rather interesting was that Eric Jang did decide to release another bunch of videos on Eve their autonomous robot doing some other tasks one of those videos was of course this right here so essentially he said we can also train on your networks to be shy by avoiding eye contact with people and you can see right here that this is of course one of their employ EMP es or someone you know working at the company trying to look at the robot and it looks completely away and I think this kind of human interaction with robot is rather fascinating as there are going to be certain Dynamics with robots that we simply haven't had yet or explored yet and this is one of the interesting things that we are going to be seeing develop as we move into an era where robots are going to be commonplace now there was also another one in which he demonstrates some more autonomy and this is where we have Eve the robot at home picking up something and then placing it on a desk now like I said whilst some people might not find this remarkably impressive I completely do because I understand how hard robotics is and just how crazy these Feats are and trust me when I say things are definitely moving pretty rapidly now there were also some other videos and before this and of course there were also some other videos that were released on their channel that were very good in terms of demonstrating the autonomous capability of this robot here you can see this robot roaming around in its fully autonomous mode and it's completely avoiding all the obstacles something that is quite hard to do and I would state that if you are going to be you know looking at robotics videos because you are someone who is interested in Ai and the future I would always recommend you take a look at the description or you know in the bottom leftand corner because a lot of the times when they do achieve autonomy it is always stated so we can assume that when they don't state that it's autonom us it's usually VR operated

Segment 3 (10:00 - 15:00)

or it's like a task specific code now there's also this video right here in which they showed us how this robot is able to autonomously go to a door and open a door something that of course humans do find to be very easy but as we do know robotics will be the last sector of AGI to based on how technology is going to be advancing now like I stated before these robots are going to be general purpose robots and of course with the reference to the first clipped where we saw them packing certain things in a warehouse style they also did release this video in which they showcase their robots working in a warehouse now I'm not sure if this is entirely autonomous it really didn't state it was more so like an advert Style video so I assume that it is partially both and partially not so this does show us that in the future when these robots are fully autonomous I can imagine them being a lot faster and doing a lot of the work that you see here in potentially some of the factories that we do know like Amazon because Amazon have been quietly deploying loads and loads of robots into their factories and we can all agree that robots are going to be much more efficient and largely going to be wanting to work hours and hours on end with no rest and that's something that traditional humans really just can't do now one of the most important things about this company is that it is actually backed by open aai so in early 2023 one actually announced it has closed its funding round of $23. 5 million led by open Ai and the company plans to use the funds to advance manufacturing of Eve in Norway and North America and increase the efforts of building the upcoming bipedo android model Neo now if you don't know what Neo is I'm going to get into that later but it's a super Advanced robot and I genuinely didn't expect the autonomous capabilities of Eve to be where it is right now but this is something that is of course good now essentially they stated that by partnering with open AI 1X wants to combine robotics with artificial intelligence to create embodied Ai and this is a serious Partnerships because other companies have been partnering together like figure and Microsoft so the humanoid robots future with artificial intelligence is clearly laid out there is a clear plan that is going on and open AI have a clear Strategic investment in this company and a partnership and it seems like this is going to be the plan platform where future AI systems whether it be AGI or agentic systems are going to be embodied in a physical nature now what they recently did as well they recently actually announced another funding round so it says here that 1X secures $800 million in a series B funding so after the initial $23. 5 million funding round they did another $100 million in funding because they needed more money to support this new model Neo now of course essentially it says here they've raised $100 million in series B funding with participation from eqt Ventures and other notable Global Investors it says the company intends to utilize the new capital to bring to Market its second generation Android Neo designed as a bipedal humanoid Neo is tailored for the everyday home assistance offering versatile support for a wide range of domestic tasks in the consumer market and the funds will also support 1X existing Enterprise clients in logistics and guarding so essentially with the $100 million that they just got they're going to be using that to build the next level robots that are going to be combined with AI backed by open Ai and of course they also do additionally see here that the funding will support their mission to produce safe and advanced Androids at a commercial scale aiming to meet Global label demands and build an abundant Society now they also tweeted this which states that we are now standing at the CP of launching our second generation biped Android Neo standing at 1... 65 M tall Neo weighs 30 kg and its exterior will be soft and made of meticulously crafted components making a new era in robotics now essentially if you don't know what Neo is because I know that I've talked about Neo and in the main part of the video we did see something called Eve essentially what you need to understand is these two models so what we have here on the left hand side is we do have Eve which is of course the robot that we do see this one is not really going to be an AI embodied robot OT but essentially it's available now and on the right hand side we have the future robotics which is of course Neo which is currently in development and they described it as your intelligent Android assistant currently in develop development and Neo is a bipedal Android with humanoid hands designed for General applications so now I want to talk a little bit about Neo because this is the robot that most people are waiting for in terms of what it's going to be because this is likely to compete with Tesla's Optimus robot which already has made significant progress now essentially I did a video

Segment 4 (15:00 - 20:00)

on this robot 6 months ago and I talked about open's first physical robot you know just shocking the entire industry because it was a giant shock to see this robot being developed now the problem is that there were severe delays because during the time when I released that video they actually spoke about Neo being announced and there being some kind of video presentation in the summer however there were delays because in that video and robotics is very hard don't think that the company's not doing their job they're doing an amazing job but robotics is really hard and essentially on the web page what they did have was they did have which is now removed how do I purchase Neo will be open for pre-orders at the end of 2023 and as far as I can tell it's actually not what they've now replaced it with is that you know safe balance and smart Neo is your intelligent Android assistant which is going to be coming soon now like I said they did of course have some delays now maybe it's some internal company stuff but I'm pretty sure it's just because of the fact that robotics is really really hard and one of the things that you might not understand is that robotics is so hard that opening ey initially gave up on Robotics and here is eliaa actually talking about why opening ey decided to give up on robotics because it was just a Monumental task and they wanted to focus on the software instead back then it really wasn't possible to continue working on robotics because there was so little data like back then if you wanted to do on robot work on robotics you needed to become a robotics company you needed to really have a giant group of people working on building robots and maintaining them and having and even then like if you only if you're going to have 100 robots it's a giant operations already but you're not going to get that much data out of so in a world where most of the progress comes from the combination of compute and data right that's where we've been where it was the combination of compute and data that drove the progress there was no path to data from robotic back in the day then you made a decision to stop working in robotic there was no path forward is there one now so I'd say that now it is possible to create a path forward but one needs to really commit to the to the task of Robotics you really need to say I'm going to build like many thousands tens of thousands hundreds of thousands of robots and somehow collect data from them and find a gradual path where the robots are doing something slightly more useful and then the data that they get from these is obtained and used to train the models they do something it's slightly more useful so you could imagine it's kind of gradual path of improvement where you build more robots they do more things you collect more data and so on but you really need to be committed to this path if you say I want to make robotics happen that's what you need to do I believe that there are companies who are thinking about such doing exactly that but I think that you need to really love robots and need to be really willing to solve all the physical and logistical problems of dealing with them it's not the same as software at all I think one could make progress in robotics today with enough motivation mhm so from that clip you can see that robotics is something that is incredibly hard but of course now you know the technology has developed to the point where that progress can be made if you want it to now essentially this is Neo so they said Neo moves like us so they can meet your demands and 1X Engineers Neo for high precision and gentle strength with arms and legs modeled after human muscle movement so they can do everything in a more natural intuitive way so now there is actually quite a lot to NE that most people need to understand because compared to last time they actually did update this page so essentially what they said here is that natural language Neo embodies artificial learning onx combines human like bodies with the minds of advanced language and autonomous movement models so you can talk with them naturally and they can do what you ask and I know that this is completely possible based on the current technology in terms of getting real time interaction with milliseconds of latency with vooo interactions with AI something that they also did include was of course that Neo's vision and movement can be quickly taken over by a remote human operator if you need support and some people might think that this is if they go Rogue but one thing that I did want to add was that I do think that this will be an interesting career opportunity and likely a emerging sector that might have a lot of demand because if we are going to be moving to a economy where Androids are everywhere and they're in our homes I like would presume that there's going to be some scenarios in which someone needs a human to take over immediately if AGI isn't completely developed by that point and then of course humans are going to have to be on call somewhere with a VR headset to be able to quickly control that robot remotely so it will be interesting to see how that field develops and matures as the robotics area expands now something that I also did find a bit strange was that they said this robot can Jog and I know that they said that it can walk jog climb stairs and navigate your space NE naturally but I don't think I've ever

Segment 5 (20:00 - 21:00)

genuinely seen a robot jog so I'm not sure if they're going to remove that or keep that or maybe the robot's going to be running like we did see with Boston Dynamics as it just completely shocked everyone but I think that is going to be something I really do want to see because I've literally never seen a robot jug just yet now there was some FAQs that I did want to add here and it said Can Neo be my personal assistant and they said whilst Neo is still in development they're being designed to be an allpurpose Android assistant to your daily life and that means that Neo will be able to take on many common tasks whether helping you around the house completing chores or running errands and then of course we had this FAQ which is how will Neo perform on general purpose label and essentially they said 1X embodied learning models combine artificial intelligence with Advanced human-like robotics to allow Neo to safely do almost anything you can and they'll serve as an extra set of hands in any context where you'll need assistance making them especially useful for the elderly or those with Mobility challenge Alles and as Neo moves they'll get smarter making them able to take on increasingly more complex tasks over time and the last one here which is of course what most people probably want to know is how does neo use Ai and essentially it says Neo will understand its environment deeper thanks to the fusion of their senses and their physical body Neo continuously learns and improves becoming smarter and more capable over time and the more you interact with them navigating life with Neo will feel natural and intuitive and Neo will understand their own environment and how to handle fragile items creating seamless collaboration and immersive human Android interaction overall hats off to the 1X team for making this remarkable feat because this is something that we haven't really seen just yet in terms of AI humanid robots and although many industry leaders are making rapid progress this is definitely a bold statement so where do you think the future is are you excited for the rise of humanid robots are you afraid do you think Terminator is just coming to fruition

Другие видео автора — TheAIGRID

Ctrl+V

Экстракт Знаний в Telegram

Экстракты и дистилляты из лучших YouTube-каналов — сразу после публикации.

Подписаться

Дайджест Экстрактов

Лучшие методички за неделю — каждый понедельник