RoboHub🤖

The future of home cleaning just landed in Shenzhen and it is walking right into your living room. 🤖🏠 @XSquareRobot and 58.com officially launched China’s first robot home service, moving embodied AI from the lab to your front door. When you book a cleaning on the 58.com app, a professional cleaner now shows up with an X Square robot partner to tag team the house. The human handles the tricky stuff that needs real judgment while the robot takes over repetitive tasks like wiping tables and tidying up surfaces. X Square is using an end to end foundation model which means the robot actually perceives and plans its own moves instead of just following a script. By testing in the messy reality of a real home, they are proving that if a robot can master a living room, it can handle almost any physical space. This pilot is part of a massive push to turn these machines into reliable partners that can actually assist in our daily lives.

@godmachinery @XSquareRobot It’s already live in Shenzhen. They’re planning to expand to more Chinese cities and bring the service into more homes soon.

@XRoboHub @XSquareRobot the human robot team approach is smart you learn more from a thousand messy living rooms than a perfect lab setup thats how you find what breaks

The future of home cleaning just landed in Shenzhen and it is walking right into your living room. 🤖🏠 @XSquareRobot and 58.com officially launched China’s first robot home service, moving embodied AI from the lab to your front door. When you book a cleaning on the 58.com app, a professional cleaner now shows up with an X Square robot partner to tag team the house. The human handles the tricky stuff that needs real judgment while the robot takes over repetitive tasks like wiping tables and tidying up surfaces. X Square is using an end to end foundation model which means the robot actually perceives and plans its own moves instead of just following a script. By testing in the messy reality of a real home, they are proving that if a robot can master a living room, it can handle almost any physical space. This pilot is part of a massive push to turn these machines into reliable partners that can actually assist in our daily lives.

@bindureddy Google will win the AI race in the West, China on Earth and SpaceX in space

@chris_j_paxton @1x_tech NEO looks sick! ! It is so cool to see you guys interacting with it in person.🤝

Shaking hands with the @1x_tech neo gamma. Its really real guys

NEO looks sick! ! It is so cool to see you guys interacting with it in person.🤝

Jensen Huang just turned the GTC 2026 stage into the ultimate "Friends" meetup for robots. 🤖 Standing in front of dozens of different robots on stage, Jensen officially declared that we have hit the "Inference Inflection Point" where AI moves from the screen into the physical world. NVIDIA is basically building the Android and AWS of the robotics era by providing the Brain, the World, and the Body in one massive tech stack. The Brain is powered by Isaac GR00T, their open source VLA foundation model, now at v1.6 and capable of full body coordination and learning complex tasks just by watching us. For the World, they are using Omniverse and the new Cosmos world model to create a digital twin playground where robots can train endlessly in simulation before ever touching the real ground. The Body is run on Jetson Thor, their dedicated edge AI platform that lets large models handle real time reasoning right on the robot itself. The robot lineup on stage was massive, including representatives from Linde, Universal Robots, XPENG, Disney Research, Boston Dynamics, Figure, Agility, 1X, and AgiBot. NVIDIA is not just selling chips anymore, they are defining the infrastructure for a projected $50 trillion robotics market. It is clear that the goal is to standardize robot development into a complete industrial AI pipeline where "Sim-to-Real" is no longer a dream but a standard operating procedure. Source: NVIDIA GTC 2026 Keynote #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #NVIDIA #GTC2026 #GR00T

Midea's six armed robot beast is already on the job, but what exactly can it do? 🤖🦾 The MIRO U humanoid is now working at Midea's laundry appliance industrial park in Wuxi. It features six coordinated arms that can lock and fasten six separate screws simultaneously with high precision. Do you think this kind of multi arm coordination will actually boost factory efficiency?

I'll get you that drink even if I have to take the whole shelf down with it. 🥤😂 As long as nobody gets hurt, watching a robot fail like this is totally fine and honestly part of the process.🤦‍♂️

If you still think robots aren't ready for real work, check out this cleaning bot using a robotic arm to actually handle chores. 🤖🧹 It doesn't just vacuum, it picks up clutter on the floor, identifies specific stains, and even swaps out its own tools for the job. It is built to navigate tight spots and manage an entire room autonomously without just bumping into furniture.

Watching North build a PC from scratch is a trip, and it is doing it fully autonomously. 🤖 From precise positioning to screwing and wire clipping, North is executing complex, contact-rich tasks using its tactile feedback. This is much more than basic automation, it is a real-world demonstration of what dexterous hands combined with tactile AI can achieve. The robot actually uses its sensors to feel the components, ensuring everything is aligned perfectly for each step of the assembly. We are seeing a major shift here, moving from robots that just follow code to machines that can actually feel their way through a task. Source: @SharpaRobotics #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #Sharpa #North #Dexterity

The AGIBOT World Challenge has officially landed at ICRA 2026 with a massive $530,000 prize pool on the line. 🤖🏆 Following the total success of the first AGIBOT World Challenge at IROS 2025, @AGIBOTofficial 2026 competition is officially coming to ICRA, the top tier of robotics conferences. This year is all about the "Robots for All" theme, moving beyond paper theories to see what these humanoids can actually handle in the real world. There are two heavy-hitter tracks: "Reasoning to Action," featuring both simulation and real-world hardware tasks like sorting, scanning, and dual-arm cooking, and a "World Model" track for physics-based video generation. This is a massive resume booster for researchers since you get to master the full VLA development loop on actual AGIBOT hardware, giving you a huge edge for top-tier lab roles or PhD apps. With open-source tools and baseline models provided, individual devs now have a real shot at tackling complex problems that used to be reserved for big-budget corporate teams. Registration and server access opened back on February 28, so make sure you hit that April 20 submission deadline. Official Challenge Info: agibot-world.com/challenge2026 Sign up here: docs.google.com/forms/d/e/1FAI… The finals are happening live on June 1 at ICRA 2026, see you guys in Vienna, Austria! #AGIBOT #AGIBOTWorldChallenge #ICRA2026 #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI

Tien Kung, the first-ever humanoid half-marathon champ, just crushed the 2026 Beijing Yizhuang test run in fully autonomous mode. 🏃‍♂️🤖

This is a robot that can actually play the piano. 🤖🎹 A research team from Stanford and Amazon FAR just dropped HandelBot, the first learning-based system to crack bimanual piano playing in the real world. Piano is a nightmare for robots because it needs millimeter-level precision and simulation alone can’t bridge that final gap. The team used simulation for the basic structure and then refined the movements using Residual RL on the actual hardware. The efficiency is wild, it mastered Twinkle Twinkle in 30 minutes and Fur Elise in 1 hour with only 100 trajectories. We are still waiting for hardware that can match a human pianist’s speed, but this leap in dexterous control is a massive win for embodied AI. Paper: arxiv.org/abs/2603.12243 Website: amberxie88.github.io/handelbot/ #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #HandelBot #Dexterity #StanfordAI

OrcaHand just dropped three new open source dexterous hands that you can 3D print and tweak for your own projects. 🤖🖐️ The star of the show is the OrcaHand Touch, which packs 83 taxels per fingertip and can detect forces as tiny as 0.1N. Having that kind of tactile feedback is a game changer for slip detection and actually handling fragile objects without crushing them. They also released a Lite version that weighs only 700g, which is perfect if you are worried about maxing out your robot arm’s payload. Everything from the source files to the print files is being opened up, plus they are promising a LEGO style assembly manual to make it dead simple. By putting high end hardware and software in everyone's hands for a low cost, they are finally making real dexterity accessible to every dev. Source: @orcahand #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #OpenSource #DexterousHand #OrcaHand

More details on the paper: x.com/Zhikai273/stat…

Humanoid tennis athletes are officially here! 🎾🤖 A team from Tsinghua University and Galbot just dropped LATENT, and it is officially teaching humanoids how to dominate on the tennis court. Reproducing high-speed athletic moves is usually a nightmare because getting "perfect" motion data for every scenario is basically impossible. LATENT cracks the code by learning from imperfect human motion fragments, using them as a structural prior to refine and compose agile, natural movements. By training a high-level policy to sample from a latent action space, the robot can now handle rapid reactions and whole-body coordination in real time. The team deployed this on a Unitree G1 and the progress was wild, on day one it couldn’t return a single serve, but by the end, the researchers actually couldn't beat it anymore. It is now sustaining stable multi-shot rallies against humans while keeping that natural athletic flow that usually takes years to master. Project: zzk273.github.io/LATENT/ Code: github.com/GalaxyGeneralR… #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #UnitreeG1 #TennisRobot #Galbot

@spacexjackie Video Via @spacexjackie

Tesla’s new 22-DOF hand is looking incredibly locked in on the Optimus v2.5 prototypes. 🤖👐 Elon has been obsessed with getting this hand right, and seeing it finally integrated into the forearm like a real human limb is a massive design win. Watching it hand out drinks right now is wild, it is just buttery smooth with zero mechanical jitter or awkwardness. Since this hardware is already crushing it in the v2.5 testing phase, you can bet the 22-degree-of-freedom setup will be the star of the v3 launch. Now that the hand is basically solved, Tesla is clearly ready on the hardware side and moving fast into the "learning" phase for complex real-world work. #Robot #Humanoid #Robotics #AI #EmbodiedAI #PhysicalAI #Tesla #Optimus #Dexterity