Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Record Set! Unitree Strikes Four Golds!🥇🥇🥇🥇 The First World Humanoid Robot Games Conclude Successfully Unitree Strikes Four Golds (1500m、400m 、100m Obstacle、4×100m Relay) The best speed achieved during the competition is 4.78 m/s (a recent internal test record exceeding 5 m/s) Congratulations as well to the independent teams competing on...

14,301,226 Aufrufe • vor 1 Jahr •via X (Twitter)

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

In just one week, Binh and I trained a full-body Unitree G1. Here's a recap: 1. Secured a Unitree G1 humanoid through a LinkedIn post 2. Deployed TWIST2 full-body teleoperation pipelines 3. Adapted TWIST2 for Zed stereo camera & collected full-body teleoperation samples (carried by Binh ) 4. Adapted & fine-tuned NVIDIA Gr00T N1.5 VLA on the TWIST2 public datasets, which I fine-tuned on an 8xNVIDIA H100 Cluster. We picked Gr00T N1.5 as it was trained with Unitree G1 embodiment data. 5. Adapted the TWIST2 codebase to stream in the actions from Gr00T via ZMQ using a co-located NVIDIA H100 for ~200ms inference latency 6. Tested the model in sim, then deployed to the real-world Unitree G1. We streamed a training sample observation to the VLA (as we didn't want to break robot in case real observations were OOD) We were the first team in the world to deploy the full TWIST2 data collection pipeline to the unitree g1 :) Much more work ahead though, which I'll work on as a side-project over the next months: 1. Exploring the various types of 'world models': video backbones, dynamics models, v-jepa-2 models. I believe these will generalize better & train much more data-efficiently than VLM backbones 2. Speeding up inference - I believe low-latency robotics inference will be a big challenge. There are many works in video diffusion which I'd like to test (e.g. SageAttention, SparseAttention, Drifting Models). Perhaps also writing custom CUDA kernels. 3. Economics of inference scaling :) What will be the compute demands as we scale inference up to millions of humanoids? Will it run on edge or on distributed 'co-located' inference clusters? These are questions I'd like to answer. Adapted TWIST2 codebase: Adapted Gr00T-N1.5 codebase: The ETH Robotics Club are doing a cool GTC Golden ticket competition with NVIDIA , so this is my submission :) The DGX Spark compute will get me a long way with initial prototyping & especially working on inference optimization for next-gen Blackwell GPUs #NVIDIAGTC #GOLDENTICKET #ETHRC

Arnie Ramesh

23,236 Aufrufe • vor 6 Monaten

Not a preplanned motion sequence. A robot deciding mid-jump what to do next. [📍 paper + demo] Researchers just showed a humanoid doing real parkour using only onboard perception. No motion script, no fixed obstacle layout. The system is called Perceptive Humanoid Parkour (PHP). Instead of memorizing a path, the robot reads depth from its cameras and continuously chooses actions. Step, vault, climb, or roll depending on what geometry appears in front of it. To make that possible, they combine three ideas: First, they stitch together human motion clips into long movement references so the robot learns fluid transitions instead of isolated tricks. Second, they train tracking policies with reinforcement learning so contacts land at the right time and the robot keeps balance during dynamic moves. Finally, everything is distilled into one perception policy that runs directly from depth input to action selection. The result on a Unitree G1: about 3 m/s vaults wall climbs up to 1.25 m nearly one minute continuous obstacle traversal adapting when obstacles move What matters is not the tricks. It is the shift in capability. Earlier humanoids executed motions. This one navigates situations. Once robots react to geometry instead of replaying trajectories, environments stop needing to be predictable. Warehouses, homes, and outdoors suddenly become the same problem. Thanks for sharing, Zhen Wu! Paper + demo: ——— Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

22,080 Aufrufe • vor 6 Monaten

Beijing hosts world’s first humanoid robot games | Ariana News The world’s inaugural humanoid robot competition is underway in Beijing, drawing more than 500 robots from 280 teams across 16 countries to compete in a uniquely futuristic sporting spectacle. The three-day event, held at the National Speed Skating Oval—once the “Ice Ribbon” of the 2022 Winter Olympics—kicked off on August 15 and runs through to Sunday August 17. The tournament features 26 events spread across athletic, performance, and scenario-based categories. Athletic challenges include sprinting, soccer, and kickboxing, while performance segments showcase robot dance routines and musical instrument displays. Real-world scenarios, such as medication sorting, cleaning tasks, and industrial material handling, are also on the agenda to test practical functionality. Organizers meanwhile emphasize the event’s role in accelerating the integration of humanoid robots into everyday life, from manufacturing and hospitality to healthcare. One Chinese official summed it up: “Every robot that participates is creating history.” The competition has yielded both triumphant strides and technical stumbling blocks. In running events, the robot H1 from Unitree Robotics claimed top honors in the 1,500-meter race, demonstrating promising agility. Yet, many robots struggled with balance, coordination, and task execution, including some collapsing mid-sprint or requiring human help to stand—underscoring the still-developing nature of embodied artificial intelligence.

Owen Gregorian

43,158 Aufrufe • vor 1 Jahr

OUR HEROES ARE HERE In a stunning display of skill, discipline, and unwavering spirit, a team of eight young Zimbabwean martial artists, guided by their dedicated master, Shihan Wilfred Mashaya, has stormed the International Combat Organisation (ICO) World Championships in Swansea, Wales, and returned home as heroes. Facing fierce competition from powerhouse nations like the United States, this small but mighty team did not just participate—they dominated. Their collective effort resulted in a historic and breathtaking haul of 43 medals for Zimbabwe, a testament to their hard work and world-class talent. The Medal Tally Speaks Volumes: 🏅 14 Gold Medals 🥈 18 Silver Medals 🥉 11 Bronze Medals “This is a proud moment for Zimbabwe,” said an elated Shihan Mashaya. “The kids worked hard, showed great discipline, and gave their best. I’m so proud of what they achieved.” This journey was about more than just winning. For many of these children, it was their first time traveling outside Zimbabwe, representing the Zimbabwe Martial Arts Kobudo/Weapon Team under the Zimbabwe Karate Union (ZKU) with pride. They competed in intense Traditional, Creative, and Freestyle Weapons events, showcasing a level of expertise that captivated the international stage. The Power of Purpose: Building Strong Children Behind the glittering medals lies a story of profound purpose. For Shihan Mashaya, martial arts is a powerful tool for social change. The months of rigorous training at the Mkushi Training Academy and Aquaview Resort at Lake Chivero were about more than preparation for a tournament; they were a sanctuary. This initiative is a direct and positive response to the scourge of drug and substance abuse plaguing the nation's youth. By instilling discipline, focus, and self-respect, Shihan Mashaya is actively building a generation of strong, resilient leaders. As the saying goes, “It is easier to build strong children than to repair broken men.” These 43 medals are 43 powerful statements against vice and apathy. Celebrating Our Champions: A Breakdown of Excellence Girls' Division Brilliance: · Dean Gwatah: 1 Silver, 3 Bronze · Anenyasha Hove: 4 Bronze · Anotida Hove: 3 Gold, 1 Silver · Thandeka Moyo: 3 Gold, 3 Silver Boys' Division Dominance: · Takudzwa Chizema: 3 Gold, 3 Silver · Wilfred Mashaya Jnr: 1 Gold, 3 Silver · Arnold Mudzengerere: 1 Gold, 2 Silver, 2 Bronze · Brian Mupindu Jnr: 1 Gold, 3 Silver, 2 Bronze Leadership by Example: Leading from the front, Shidoshi Wilfred Mashaya was also crowned Kobudo World Champion, securing 2 Gold and 2 Silver medals in the black belt categories, proving that the master’s skill matches his inspirational leadership.

Nick Mangwana

16,580 Aufrufe • vor 10 Monaten

The next two months put $IREN's entire pivot on one milestone. Horizon 1, its first GB300 super cluster at Childress, is targeted to hand off to $MSFT in Q3, roughly July through September, the opening delivery under the five year, $9.7 billion Microsoft contract and the point where that contract starts turning into revenue. For a year the company has been buying the hard things. Power, land, financing. Now it has to turn them into delivered compute. That is the whole test. "The world is structurally short compute, and the bottleneck is delivered data center and GPU capacity," said Daniel Roberts, Co-Founder and Co-CEO of $IREN. Horizon 1 is him putting that to the proof. Capacity nobody can energize is worthless. Capacity handed to a hyperscaler on schedule is the business. The rest of the window fills in around it. The Mirantis acquisition, signed in May, is pending close and adds the software layer to run the fleet. $3.1 billion of ARR sits under contract against a $4.4 billion target, so there is room for another customer, and management is openly chasing one. The $3.65 billion GPU financing that closed June 1 already funds most of the $MSFT hardware, with $NVDA and $DELL on the supply side. Sweetwater 1, energized in May, keeps ramping power behind all of it. The full year FY2026 results that put real numbers on all of this land just past the window, late August into September. The catalysts here are operational first, reported second. What makes that timeline credible is the record behind it. $IREN hit 50 EH/s on the schedule it set, energized Sweetwater 1 on schedule, and has Horizon 1-4 tracking for year-end. This is a team that keeps turning secured power into online capacity on time, and each build makes the next one faster. How many names in this AI buildout are actually delivering capacity on schedule, not just announcing it? It's not a sprint, it's a marathon.

Patient Investor

105,170 Aufrufe • vor 2 Monaten