Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Using only box-forwarding speed as the reward, our Stackelberg PPO automatically evolves robots with arms for pushing and legs for moving. The key idea is a novel game-theoretic view of structure–control co-design, yielding more effective optimization and dramatically better designs. Come see our poster at ICLR 2026 on Apr... show more

Jürgen Schmidhuber

223,255 subscribers

51,765 views • 5 months ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Model-Free Reinforcement Learning (MFRL) has been alluring, especially with supercharged compute with physics on GPU. However, the methods use 0-th order gradients, and are often not the best optimizers. Can we do better than PPO in continuous control for robotics? Turns out yes! 🥳 tl;dr: Faster, better RL than PPO in continuous control 💪 The answer lies in using more information from the simulation. We are juicing the simulation on GPU as it is, why not use it for gradients as well? This has been a driving question in a series of our works. We first studied this problem in ICLR 2022 paper on Short Horizon Actor Critic Naive gradient based methods are stuck in local minima and have exploding/vanishing gradients. SHAC solved this problem truncated rollouts and model based value estimation, where the model is Differentiable Sim. This boosted sample efficiency and wall-clock time immensely especially in high dimensional systems such as humanoids Yet, given enough compute PPO often caught up. Our follow up paper on on Adaptive Horizon Actor Critic at ICML 2024 discovers the cause and provides a fix. However, we find that even when given ground-truth dynamics, not all gradients are useful due to sample error. 1st-Order Model-Based Reinforcement Learning methods employing differentiable simulation provide gradients with reduced variance but are susceptible to bias in scenarios involving stiff dynamics, such as physical contact. We find that back-propagating through contact and long trajectories drastically reduces gradient accuracy. Using this insight, we propose AHAC to dynamically adapt its roll-out horizon to avoid differentiating through stiff contact. AHAC is a first-order model-based RL algorithm that learns high-dimensional tasks in minutes (wall clock) and outperforms PPO by 40%, even in the limit of data provided to PPO. This work is led by Ignat Georgiev alongside Krishnan Srinivasan, Jie Xu, Eric Heiden and ample assistance from warp team at NVIDIA Robotics (Miles Macklin)

Model-Free Reinforcement Learning (MFRL) has been alluring, especially with supercharged compute with physics on GPU. However, the methods use 0-th order gradients, and are often not the best optimizers. Can we do better than PPO in continuous control for robotics? Turns out yes! 🥳 tl;dr: Faster, better RL than PPO in continuous control 💪 The answer lies in using more information from the simulation. We are juicing the simulation on GPU as it is, why not use it for gradients as well? This has been a driving question in a series of our works. We first studied this problem in ICLR 2022 paper on Short Horizon Actor Critic Naive gradient based methods are stuck in local minima and have exploding/vanishing gradients. SHAC solved this problem truncated rollouts and model based value estimation, where the model is Differentiable Sim. This boosted sample efficiency and wall-clock time immensely especially in high dimensional systems such as humanoids Yet, given enough compute PPO often caught up. Our follow up paper on on Adaptive Horizon Actor Critic at ICML 2024 discovers the cause and provides a fix. However, we find that even when given ground-truth dynamics, not all gradients are useful due to sample error. 1st-Order Model-Based Reinforcement Learning methods employing differentiable simulation provide gradients with reduced variance but are susceptible to bias in scenarios involving stiff dynamics, such as physical contact. We find that back-propagating through contact and long trajectories drastically reduces gradient accuracy. Using this insight, we propose AHAC to dynamically adapt its roll-out horizon to avoid differentiating through stiff contact. AHAC is a first-order model-based RL algorithm that learns high-dimensional tasks in minutes (wall clock) and outperforms PPO by 40%, even in the limit of data provided to PPO. This work is led by Ignat Georgiev alongside Krishnan Srinivasan, Jie Xu, Eric Heiden and ample assistance from warp team at NVIDIA Robotics (Miles Macklin)

Animesh Garg

52,308 views • 2 years ago

With 3 more sales today, we've now hit the milestone and officially burned 20.2% of the entire supply of NODESTR burned in 6.5 months. That comes out to an average of more than 0.1% of supply burned EVERY DAY since we began. Even the most well known flywheel DAT on ETH only burned 6.3% before death spiraling, but we structured our project for long term success, and we remain committed to our long term goals. Some of our initial project design choices weren't popular at the time, but our ironclad foundation is being proven as correct as more time goes on. Our project has always been a long term asymmetric bet on the providence of the most rare digital assets on the original chain where NFT's were originally born. We don't know how long the journey will take, but we will continue to plow forward relentlessly.

With 3 more sales today, we've now hit the milestone and officially burned 20.2% of the entire supply of NODESTR burned in 6.5 months. That comes out to an average of more than 0.1% of supply burned EVERY DAY since we began. Even the most well known flywheel DAT on ETH only burned 6.3% before death spiraling, but we structured our project for long term success, and we remain committed to our long term goals. Some of our initial project design choices weren't popular at the time, but our ironclad foundation is being proven as correct as more time goes on. Our project has always been a long term asymmetric bet on the providence of the most rare digital assets on the original chain where NFT's were originally born. We don't know how long the journey will take, but we will continue to plow forward relentlessly.

NodeStrategy

10,526 views • 5 months ago

Welcome to the Dwayne “the Rock” Johnson Football Locker Room at the one and only, University of Miami 🙌🏾 THE U At Miami, I had the privilege of playing with some of the greatest players to ever play the game. Hall of Famers. And more importantly, WE had the honor of creating a brotherhood that was forged in our daily discipline ~ to outwork our competition and ALWAYS be the hardest workers in the room. Lifers. This gift is my way of showing love and gratitude to a football program that helped shape and define who I am today as a man. Ups and downs and everything in between. Best of luck this season to my brother & Head Coach Mario Cristobal & his coaching staff. And to our program’s anchor - our players, No team in the country will outwork you. It’s our obsession and our appetite for disruption. See you at the top. DRJ #brotherhood #theU University of Miami Miami Hurricanes Football

Welcome to the Dwayne “the Rock” Johnson Football Locker Room at the one and only, University of Miami 🙌🏾 THE U At Miami, I had the privilege of playing with some of the greatest players to ever play the game. Hall of Famers. And more importantly, WE had the honor of creating a brotherhood that was forged in our daily discipline ~ to outwork our competition and ALWAYS be the hardest workers in the room. Lifers. This gift is my way of showing love and gratitude to a football program that helped shape and define who I am today as a man. Ups and downs and everything in between. Best of luck this season to my brother & Head Coach Mario Cristobal & his coaching staff. And to our program’s anchor - our players, No team in the country will outwork you. It’s our obsession and our appetite for disruption. See you at the top. DRJ #brotherhood #theU University of Miami Miami Hurricanes Football

Dwayne Johnson

3,058,538 views • 3 years ago

𝗥𝗼𝗯𝗼𝘁𝘀 𝗱𝗼𝗻’𝘁 𝗻𝗲𝗲𝗱 𝗺𝗼𝗿𝗲 𝗱𝗲𝗺𝗼𝗻𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻𝘀. 𝗧𝗵𝗲𝘆 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗹𝗲𝗮𝗿𝗻 𝗳𝗿𝗼𝗺 𝗳𝗮𝗶𝗹𝘂𝗿𝗲 — 𝗮𝗳𝘁𝗲𝗿 𝘄𝗮𝘁𝗰𝗵𝗶𝗻𝗴 𝗵𝘂𝗺𝗮𝗻𝘀. Most robot learning systems assume failure is the end of learning. In our new work, we study whether robots can improve after deployment by learning from their own failures, without any human intervention, teleoperation, or corrective labels. The key idea is simple: human videos contain structure about how the world works. We use them to learn cross-embodiment representations of action, dynamics, and value, enabling a shared predictive space between human behavior and robot experience. This allows a new learning loop: 👉 pretrain on human videos 👉 deploy robot policy 👉 observe failures 👉 reinterpret failures using human priors 👉 improve autonomously We evaluate this across 7 real-world manipulation tasks, showing: 📈 40% → 81% success rate 🏆 Strong improvements over π0.6 RECAP and RISE ✔️ Zero human intervention during post-deployment improvement 🧬 Generalizes across robot embodiments and policy backbones A key finding is that explicit failure repair significantly outperforms failure reweighting, yielding substantially larger gains under identical data conditions (+25 pts vs +5 pts on the same π0.5 base policy). Overall, the results suggest a shift in how we think about robot learning: Human videos are not only for pretraining policies. They can provide the structure needed for continual self-improvement after deployment. 📄 Paper: 🌐 Project: I am grateful for working with the fantastic leads Hanzhi Chen and Anran Zhang, and our collaborators Simon Schaefer, Kejia Chen, Shi Chen, Daniel Cremers. Special thanks to Stefan Leutenegger for co-advising this project with me. ETH Zürich TU München Microsoft Check out Hanzhi's 🧵 for more details

𝗥𝗼𝗯𝗼𝘁𝘀 𝗱𝗼𝗻’𝘁 𝗻𝗲𝗲𝗱 𝗺𝗼𝗿𝗲 𝗱𝗲𝗺𝗼𝗻𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻𝘀. 𝗧𝗵𝗲𝘆 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗹𝗲𝗮𝗿𝗻 𝗳𝗿𝗼𝗺 𝗳𝗮𝗶𝗹𝘂𝗿𝗲 — 𝗮𝗳𝘁𝗲𝗿 𝘄𝗮𝘁𝗰𝗵𝗶𝗻𝗴 𝗵𝘂𝗺𝗮𝗻𝘀. Most robot learning systems assume failure is the end of learning. In our new work, we study whether robots can improve after deployment by learning from their own failures, without any human intervention, teleoperation, or corrective labels. The key idea is simple: human videos contain structure about how the world works. We use them to learn cross-embodiment representations of action, dynamics, and value, enabling a shared predictive space between human behavior and robot experience. This allows a new learning loop: 👉 pretrain on human videos 👉 deploy robot policy 👉 observe failures 👉 reinterpret failures using human priors 👉 improve autonomously We evaluate this across 7 real-world manipulation tasks, showing: 📈 40% → 81% success rate 🏆 Strong improvements over π0.6 RECAP and RISE ✔️ Zero human intervention during post-deployment improvement 🧬 Generalizes across robot embodiments and policy backbones A key finding is that explicit failure repair significantly outperforms failure reweighting, yielding substantially larger gains under identical data conditions (+25 pts vs +5 pts on the same π0.5 base policy). Overall, the results suggest a shift in how we think about robot learning: Human videos are not only for pretraining policies. They can provide the structure needed for continual self-improvement after deployment. 📄 Paper: 🌐 Project: I am grateful for working with the fantastic leads Hanzhi Chen and Anran Zhang, and our collaborators Simon Schaefer, Kejia Chen, Shi Chen, Daniel Cremers. Special thanks to Stefan Leutenegger for co-advising this project with me. ETH Zürich TU München Microsoft Check out Hanzhi's 🧵 for more details

Oier Mees

12,732 views • 3 months ago

3D Gaussian Splatting for Real-Time Radiance Field Rendering paper page: Radiance Field methods have recently revolutionized novel-view synthesis of scenes captured with multiple photos or videos. However, achieving high visual quality still requires neural networks that are costly to train and render, while recent faster methods inevitably trade off speed for quality. For unbounded and complete scenes (rather than isolated objects) and 1080p resolution rendering, no current method can achieve real-time display rates. We introduce three key elements that allow us to achieve state-of-the-art visual quality while maintaining competitive training times and importantly allow high-quality real-time (>= 30 fps) novel-view synthesis at 1080p resolution. First, starting from sparse points produced during camera calibration, we represent the scene with 3D Gaussians that preserve desirable properties of continuous volumetric radiance fields for scene optimization while avoiding unnecessary computation in empty space; Second, we perform interleaved optimization/density control of the 3D Gaussians, notably optimizing anisotropic covariance to achieve an accurate representation of the scene; Third, we develop a fast visibility-aware rendering algorithm that supports anisotropic splatting and both accelerates training and allows realtime rendering. We demonstrate state-of-the-art visual quality and real-time rendering on several established datasets.

3D Gaussian Splatting for Real-Time Radiance Field Rendering paper page: Radiance Field methods have recently revolutionized novel-view synthesis of scenes captured with multiple photos or videos. However, achieving high visual quality still requires neural networks that are costly to train and render, while recent faster methods inevitably trade off speed for quality. For unbounded and complete scenes (rather than isolated objects) and 1080p resolution rendering, no current method can achieve real-time display rates. We introduce three key elements that allow us to achieve state-of-the-art visual quality while maintaining competitive training times and importantly allow high-quality real-time (>= 30 fps) novel-view synthesis at 1080p resolution. First, starting from sparse points produced during camera calibration, we represent the scene with 3D Gaussians that preserve desirable properties of continuous volumetric radiance fields for scene optimization while avoiding unnecessary computation in empty space; Second, we perform interleaved optimization/density control of the 3D Gaussians, notably optimizing anisotropic covariance to achieve an accurate representation of the scene; Third, we develop a fast visibility-aware rendering algorithm that supports anisotropic splatting and both accelerates training and allows realtime rendering. We demonstrate state-of-the-art visual quality and real-time rendering on several established datasets.

AK

634,252 views • 3 years ago

Disappointed with your ICLR paper being rejected? Ten years ago today, Sergey and I finished training some of the first end-to-end neutral nets for robot control 🤖 We submitted the paper to RSS on January 23, 2015. It was rejected for being "incremental" and "unlikely to have much impact" Our resubmission to NeurIPS was also rejected It now has >4,000 citations (and more importantly, end-to-end training is widely accepted!) It's also cool to think about what's changed and what's the same -- - The network was 92k parameters and trained on ~15 minutes of data - The code was a combination of matlab, caffe, ROS, a custom CUDA kernel for speed, and a low-level 20 Hz controller in C++, all talking to each other. ROS+matlab was as bad as it sounds. - We pre-trained the encoder and did inference off-board on a workstation with a larger GPU. - We were paranoid about varying lighting messing up the network, so we did all the experiments after sunset (so long nights running experiments on the robot past 3 am) Now, we have manipulation policies that are far more dextrous, far more generalizable, and maybe on the cusp of breaking into the real world. :) (the paper:

Disappointed with your ICLR paper being rejected? Ten years ago today, Sergey and I finished training some of the first end-to-end neutral nets for robot control 🤖 We submitted the paper to RSS on January 23, 2015. It was rejected for being "incremental" and "unlikely to have much impact" Our resubmission to NeurIPS was also rejected It now has >4,000 citations (and more importantly, end-to-end training is widely accepted!) It's also cool to think about what's changed and what's the same -- - The network was 92k parameters and trained on ~15 minutes of data - The code was a combination of matlab, caffe, ROS, a custom CUDA kernel for speed, and a low-level 20 Hz controller in C++, all talking to each other. ROS+matlab was as bad as it sounds. - We pre-trained the encoder and did inference off-board on a workstation with a larger GPU. - We were paranoid about varying lighting messing up the network, so we did all the experiments after sunset (so long nights running experiments on the robot past 3 am) Now, we have manipulation policies that are far more dextrous, far more generalizable, and maybe on the cusp of breaking into the real world. :) (the paper:

Chelsea Finn

169,471 views • 1 year ago

A look back at our mission to do /more. 🔥🛠️ Months ago, we entered a new growth phase for xLaunchpad, driven by a simple commitment: to do more. More exciting startups, more platform improvements, and a deeper, more meaningful connection with our community. /new We’ve introduced multiple initiatives to make participation in xLaunchpad more accessible, while still keeping EGLD stakers at the core of the experience: 🔸 Challenges Portal – Built in partnership with ᕈulsar Money, this platform allows users to earn lottery tickets by completing simple tasks. 🔸 Creators Program – A reward system for users who create valuable content about xLaunchpad and its projects. 🔸 xLaunchpad Users – Loyal participants from previous launches now have a chance to earn lottery tickets. 🔸 Project Users – Projects launching on xLaunchpad can distribute lottery tickets to eligible users based on set criteria. 🔸 EGLD Holders – Tickets can now be earned simply by holding EGLD in a wallet. Beyond these initiatives, we’ve also strengthened relationships with key community members, empowering them to write educational content about upcoming startups with full creative control on their side. And of course, we’ve introduced more projects and focused on accelerating launch timelines. /better The platform and lottery system have undergone continuous improvements, such as making participation more accessible as detailed above: 🔸 Faster EGLD Reclaims - Users now get their EGLD back more quickly after participation. 🔸 Improved Communication – Blog & X Articles covering both general topics and project-specific updates. Increased community discussions and more structured feedback collection. 🔸 Website Enhancements - Various UI/UX tweaks, with /more to come 👀. 🔸 Integrating AshSwap - Users could buy lottery tickets for the last project with any tokens. 🔸 KYC Improvements - Streamlined processes, with /more to come 👀. Additionally, we’ve successfully co-launched projects with other launchpads and supported startups in securing partnerships, listings, business deals, and legal guidance. xLaunchpad remains one of the most compliant launchpads, ensuring projects align with MiCA regulations and broader industry standards. /next Looking ahead, our main focus remains on the product and the community. Every improvement is guided by real user interactions and feedback, ensuring xLaunchpad continues to evolve. At its core, our mission revolves around two key goals: 1️⃣ Bringing exciting startups to life 2️⃣ Providing users with investment opportunities on the best terms For xLaunchpad and its community to thrive together, it’s essential to recognize these two pillars while understanding the platform’s bigger purpose: making something possible that wasn’t before - allowing retail users to participate in the early launch of promising projects and benefit through multiple avenues. From investments to early adoption, and from unlocking key benefits to shaping young startups. We cannot stress enough how much effort the team puts into achieving these key goals, and we’re grateful to have you by our side on this mission. /more

A look back at our mission to do /more. 🔥🛠️ Months ago, we entered a new growth phase for xLaunchpad, driven by a simple commitment: to do more. More exciting startups, more platform improvements, and a deeper, more meaningful connection with our community. /new We’ve introduced multiple initiatives to make participation in xLaunchpad more accessible, while still keeping EGLD stakers at the core of the experience: 🔸 Challenges Portal – Built in partnership with ᕈulsar Money, this platform allows users to earn lottery tickets by completing simple tasks. 🔸 Creators Program – A reward system for users who create valuable content about xLaunchpad and its projects. 🔸 xLaunchpad Users – Loyal participants from previous launches now have a chance to earn lottery tickets. 🔸 Project Users – Projects launching on xLaunchpad can distribute lottery tickets to eligible users based on set criteria. 🔸 EGLD Holders – Tickets can now be earned simply by holding EGLD in a wallet. Beyond these initiatives, we’ve also strengthened relationships with key community members, empowering them to write educational content about upcoming startups with full creative control on their side. And of course, we’ve introduced more projects and focused on accelerating launch timelines. /better The platform and lottery system have undergone continuous improvements, such as making participation more accessible as detailed above: 🔸 Faster EGLD Reclaims - Users now get their EGLD back more quickly after participation. 🔸 Improved Communication – Blog & X Articles covering both general topics and project-specific updates. Increased community discussions and more structured feedback collection. 🔸 Website Enhancements - Various UI/UX tweaks, with /more to come 👀. 🔸 Integrating AshSwap - Users could buy lottery tickets for the last project with any tokens. 🔸 KYC Improvements - Streamlined processes, with /more to come 👀. Additionally, we’ve successfully co-launched projects with other launchpads and supported startups in securing partnerships, listings, business deals, and legal guidance. xLaunchpad remains one of the most compliant launchpads, ensuring projects align with MiCA regulations and broader industry standards. /next Looking ahead, our main focus remains on the product and the community. Every improvement is guided by real user interactions and feedback, ensuring xLaunchpad continues to evolve. At its core, our mission revolves around two key goals: 1️⃣ Bringing exciting startups to life 2️⃣ Providing users with investment opportunities on the best terms For xLaunchpad and its community to thrive together, it’s essential to recognize these two pillars while understanding the platform’s bigger purpose: making something possible that wasn’t before - allowing retail users to participate in the early launch of promising projects and benefit through multiple avenues. From investments to early adoption, and from unlocking key benefits to shaping young startups. We cannot stress enough how much effort the team puts into achieving these key goals, and we’re grateful to have you by our side on this mission. /more

xLaunchpad

22,927 views • 1 year ago

The renovation of Belleair CC’s East Course involves moving some seriously mature trees. The process and planning for moving these trees is very thorough. The result- a completely renovated golf course with a seasoned feel to it. Unlike the restoration of our West course, the East is only restoring about a third of Ross’ design. Why not restore it all? It was an impossible task- back in the 1980s our current clubhouse was built over the original driving range. At that time we built a new driving range and lost two par threes and the opening holes of the East. Around that same time the club needed a irrigation pond for the 36 holes, which ate up a good portion of Ross’ East Course design on the other side of the road. The result of that redesign was a course which had 6 subpar golf holes. This renovation will create a better golf course designed to challenge all skill levels. I am very excited to have a better East Course!

The renovation of Belleair CC’s East Course involves moving some seriously mature trees. The process and planning for moving these trees is very thorough. The result- a completely renovated golf course with a seasoned feel to it. Unlike the restoration of our West course, the East is only restoring about a third of Ross’ design. Why not restore it all? It was an impossible task- back in the 1980s our current clubhouse was built over the original driving range. At that time we built a new driving range and lost two par threes and the opening holes of the East. Around that same time the club needed a irrigation pond for the 36 holes, which ate up a good portion of Ross’ East Course design on the other side of the road. The result of that redesign was a course which had 6 subpar golf holes. This renovation will create a better golf course designed to challenge all skill levels. I am very excited to have a better East Course!

Society of Golf Historians

15,039 views • 5 months ago

This isn’t a 3D model file. It’s 100% code Our world-famous Mars Cat avatar has been reconstructed entirely as Three.js source code No GLB at runtime. No model download. No textures or UV atlas. Just code, rendered directly in the browser It was created with IMG2THREEJS by Nick. An incredible open-source project with over 15,000 stars on GitHub And our Mars Cat is now featured on the home page gallery See it live below ↓. Grab the code. Add it to your game New to Three.js? It’s one of the most popular open-source JavaScript libraries for building interactive 3D experiences and games on the web

Jay B 😻🚀

15,044 views • 1 month ago

Saronic has been awarded a production OT under the U.S. Navy's MUSV Marketplace for Marauder, our 180-foot dual-use autonomous ship — following a rigorous at-sea evaluation that tested the vessel’s capabilities. The trials included autonomously tracking and maneuvering around a variety of vessels, plus a battery of passive-perception tests — requiring Marauder to detect, track, and follow other vessels using only its passive onboard sensors, with no active radar. It's one of the hardest problems in autonomous maritime systems, and Marauder ran every scenario back-to-back on the same platform now under contract with the Navy. "Marauder went from initial design to on-the-water in under a year, and this evaluation demonstrated that the platform holds up under the Navy's own toughest testing," said Dino Mavrookas, CEO and Co-Founder. "We've now flipped our third Marauder hull, with several more moving through the production line behind it. We can build at that pace because we designed the autonomy, software, hardware, and the production process side-by-side, as one system, rather than bolting them together after the fact. That's what it means to be a vertically integrated shipbuilder, and it's why we've been able to build and deliver at speed and scale."

Saronic has been awarded a production OT under the U.S. Navy's MUSV Marketplace for Marauder, our 180-foot dual-use autonomous ship — following a rigorous at-sea evaluation that tested the vessel’s capabilities. The trials included autonomously tracking and maneuvering around a variety of vessels, plus a battery of passive-perception tests — requiring Marauder to detect, track, and follow other vessels using only its passive onboard sensors, with no active radar. It's one of the hardest problems in autonomous maritime systems, and Marauder ran every scenario back-to-back on the same platform now under contract with the Navy. "Marauder went from initial design to on-the-water in under a year, and this evaluation demonstrated that the platform holds up under the Navy's own toughest testing," said Dino Mavrookas, CEO and Co-Founder. "We've now flipped our third Marauder hull, with several more moving through the production line behind it. We can build at that pace because we designed the autonomy, software, hardware, and the production process side-by-side, as one system, rather than bolting them together after the fact. That's what it means to be a vertically integrated shipbuilder, and it's why we've been able to build and deliver at speed and scale."

Saronic

230,552 views • 13 days ago

Robots can now reconstruct 3D scenes in real time from a single RGB camera. [📍 Projects page + paper] No depth sensor. No retraining. 30 FPS. Researchers at the Imperial College London introduced KV-Tracker, a training-free method that makes heavy models like π³ and Depth Anything 3 fast enough for real-time tracking. The idea is simple. These models use global self-attention, which is powerful but computationally expensive. KV-Tracker caches the key and value pairs from selected keyframes and reuses them for new frames. That cache becomes an implicit scene representation. Result: • Up to 30 FPS • 10 to 15x speedup • Accurate 6-DoF tracking on benchmarks like TUM RGB-D and 7-Scenes • Works with monocular RGB only It also supports object-level tracking with masks and allows saving the KV-cache for later reuse. For robotics, this reduces hardware constraints and moves real-time 3D perception closer to practical deployment. Credit to Marwan Taher (Marwan Taher) at Imperial’s Dyson Robotics Lab and many others who contributed to this! 📍 Save projects page + paper for later: Video: ——- if it matters in AI or Robotics you'll read it here first:

Robots can now reconstruct 3D scenes in real time from a single RGB camera. [📍 Projects page + paper] No depth sensor. No retraining. 30 FPS. Researchers at the Imperial College London introduced KV-Tracker, a training-free method that makes heavy models like π³ and Depth Anything 3 fast enough for real-time tracking. The idea is simple. These models use global self-attention, which is powerful but computationally expensive. KV-Tracker caches the key and value pairs from selected keyframes and reuses them for new frames. That cache becomes an implicit scene representation. Result: • Up to 30 FPS • 10 to 15x speedup • Accurate 6-DoF tracking on benchmarks like TUM RGB-D and 7-Scenes • Works with monocular RGB only It also supports object-level tracking with masks and allows saving the KV-cache for later reuse. For robotics, this reduces hardware constraints and moves real-time 3D perception closer to practical deployment. Credit to Marwan Taher (Marwan Taher) at Imperial’s Dyson Robotics Lab and many others who contributed to this! 📍 Save projects page + paper for later: Video: ——- if it matters in AI or Robotics you'll read it here first:

Ilir Aliu

53,992 views • 6 months ago

It’s hard to believe we're on the final day of our campaign to save the farm and keep our special Esther Magic in Campbellville with Operation Angels. In just two short months, we've raised over half a million dollars to help bring this dream to life. Although we haven't hit our goal yet, what we've achieved is incredible. With over 6,100 contributors from 39 countries totaling over $560,000, this is a global affair in typical Esther fashion. This project is designed to generate revenue to support sanctuaries and rescue organizations and serve as a model for revenue-generating programs at existing facilities. What sets us apart is our approach to drawing revenue from a new demographic to ease the pressure on current donors and an educational program to better equip animal organizations for the challenges they face. This project represents everything Esther stood for: kindness, compassion, inclusion, and support. I can't think of a better way to honor her and the community of like-minded people who make our little corner of the internet unique. The campaign will remain active until 4 am ET on August 1st, so we have less than 17 hours to give it our best and finish strong. To anyone who hasn't contributed, I hope you’ll consider doing so today because we could certainly use your help. We may not reach our goal, and that’s okay. Don’t be discouraged. It just means we keep going because come hell or high water, we’re going back to Campbellville. Thank you to everyone who has contributed so far and to everyone who has liked, shared, and commented on campaign posts these last two months. You have all played a critical role, and I am eternally grateful for your support and encouragement. It is an honor and a privilege to be part of this project with all of you. It represents the end of an era and the start of a new and exciting one. We’re going to do amazing things together, and I can't wait to get back to work. See you all live tonight for our final campaign Garden Chat at 6:30 PM ET. xoxo Steve.

It’s hard to believe we're on the final day of our campaign to save the farm and keep our special Esther Magic in Campbellville with Operation Angels. In just two short months, we've raised over half a million dollars to help bring this dream to life. Although we haven't hit our goal yet, what we've achieved is incredible. With over 6,100 contributors from 39 countries totaling over $560,000, this is a global affair in typical Esther fashion. This project is designed to generate revenue to support sanctuaries and rescue organizations and serve as a model for revenue-generating programs at existing facilities. What sets us apart is our approach to drawing revenue from a new demographic to ease the pressure on current donors and an educational program to better equip animal organizations for the challenges they face. This project represents everything Esther stood for: kindness, compassion, inclusion, and support. I can't think of a better way to honor her and the community of like-minded people who make our little corner of the internet unique. The campaign will remain active until 4 am ET on August 1st, so we have less than 17 hours to give it our best and finish strong. To anyone who hasn't contributed, I hope you’ll consider doing so today because we could certainly use your help. We may not reach our goal, and that’s okay. Don’t be discouraged. It just means we keep going because come hell or high water, we’re going back to Campbellville. Thank you to everyone who has contributed so far and to everyone who has liked, shared, and commented on campaign posts these last two months. You have all played a critical role, and I am eternally grateful for your support and encouragement. It is an honor and a privilege to be part of this project with all of you. It represents the end of an era and the start of a new and exciting one. We’re going to do amazing things together, and I can't wait to get back to work. See you all live tonight for our final campaign Garden Chat at 6:30 PM ET. xoxo Steve.

Esther TheWonder Pig

11,928 views • 2 years ago

REO Trucks is proud to announce frankstephenson design, as its design partner for the new Runabout lineup. Over four decades, Frank Stephenson has designed or led the design of the modern MINI, the first BMW X5, the Fiat 500, the Ferrari F430, the Maserati MC12 and the McLaren P1. frankstephenson design is working with REO on all three Runabout models: the T4X work truck, the T4C crew cab and the S4C SUV which are being designed around reliability and right to repair. Frank Stephenson, founder of frankstephenson design, said: “We’re proud to be working with REO, a brand with more than a century of American history behind it. It isn’t often you get the chance to help bring a name like this back, and the team of visionaries behind REO has a very clear idea of what these trucks need to be: simple, honest and built to last. That’s a brilliant brief for any designer. Our team is really enjoying shaping all three Runabout models, and we can’t wait to start showing people the work.” Teasers will be released over the coming weeks, building to a digital reveal by November, with the first physical vehicle to follow in 2027. Reservations remain open at for a fully refundable $25.

REO Industries

104,694 views • 21 days ago

Thank you all so much for the incredible support over the past day towards our editions reveal! Because of you, we just ticked over 1.4 MILLION wishlists 🥳 You have been asking lots of important questions, so here is a quick rundown to keep you informed on all things MOUSE: P.I. For Hire 🐭 🖥️ The game is still launching digitally on April 16, only physical editions will release on July 10, due to longer production timelines 🎮 Switch 2 physical editions are full game-on cartridges, not game-key cards 🛒 Pre-purchases are currently only live for Steam and will not be available for PS5 & Switch 2, but we are exploring pre-orders for Xbox Series X|S 🕹️ We will not be doing a demo, as the team is fully focused on the launch build 💿 The game will not see a release on PS4, Xbox One and Switch 1 on April 16, and details for previous-gen console releases will be revealed at a later stage 💚 The game will not be on Xbox Game Pass and physical Standard Editions are not currently planned, but it will support Xbox Play Anywhere 🕵️‍♂️ The Story DLC will come to PC (Steam), PS5, Switch 2 and Xbox Series, and it will be a brand new expansion, separate from the main story in the base game 🗺️ Check your local retailers in the coming weeks for physical edition listings to populate and take a look at region availability on our website More information here:

MOUSE: P.I. For Hire

150,682 views • 6 months ago

A Letter to Our Community: The Road Ahead for Robotics To our Community and Partners, As we step into 2026, our mission at Axis is clearer than ever: Constructing the definitive End-to-End Scaling Layer for Robotics. Our goal is to accelerate the transfer of diverse human intelligence into Robotics General Intelligence (RGI). By owning the critical path of intelligence creation, we are turning the physical limitations of robotics into a scalable, software-driven future. Here is our strategic outlook and roadmap for the year ahead. The Core Thesis: Simulation is the Only Way Out The path to RGI is currently blocked by Data Scarcity, Generalization Fragility, and Hardware Fragmentation. At Axis, we believe Simulation is the only way out. Our Simulation Data Platform and Data Augmentation Engine transform raw data into "Synthetic Gold". Backed by academic milestones like Roboverse, Skill Blending, and GraspVLA, we have proven that pure simulation can achieve the generalization required for the real world. We don’t just collect data; we architect it. The Engine: Why Crypto? We believe RGI should come from all, not a few. Crypto is not just a feature; it is the primitive that powers our entire ecosystem flywheel: - Incentive Mechanism: Democratizing contribution and rewarding the trainers and developers. - Assetization: Turning proprietary data and refined models into liquid, ownable assets. - Verifiable Workflow: We are opening the "Black Box" of AI. By bringing total transparency to the Task Generation → Data Collection → Model Training pipeline, we ensure every byte of intelligence is verifiable, traceable, and secure. 2026 Strategic Deliverables This year, we are committed to delivering three foundational pillars: - The World's Largest Training Dataset for Robots: A robot training set—diverse, high-quality interaction data at an unprecedented scale. - A Robotics Foundation Model: A universal robotic brain trained on our pure simulation and synthetic data, capable of robust cross-embodiment transfer and open-world adaptability. - Evolvable Robot Hardware: Robots deployed with Axis models that autonomously evolve through continuous interaction, turning every deployment into a self-improving node within our RGI network. The Ultimate Vision We are building more than models; we are architecting the Distributed Machine Economy. A future where every dataset, model, and robotic embodiment is a verifiable asset in a global, autonomous network. Thank you for building the future of intelligence with us✌️📷

A Letter to Our Community: The Road Ahead for Robotics To our Community and Partners, As we step into 2026, our mission at Axis is clearer than ever: Constructing the definitive End-to-End Scaling Layer for Robotics. Our goal is to accelerate the transfer of diverse human intelligence into Robotics General Intelligence (RGI). By owning the critical path of intelligence creation, we are turning the physical limitations of robotics into a scalable, software-driven future. Here is our strategic outlook and roadmap for the year ahead. The Core Thesis: Simulation is the Only Way Out The path to RGI is currently blocked by Data Scarcity, Generalization Fragility, and Hardware Fragmentation. At Axis, we believe Simulation is the only way out. Our Simulation Data Platform and Data Augmentation Engine transform raw data into "Synthetic Gold". Backed by academic milestones like Roboverse, Skill Blending, and GraspVLA, we have proven that pure simulation can achieve the generalization required for the real world. We don’t just collect data; we architect it. The Engine: Why Crypto? We believe RGI should come from all, not a few. Crypto is not just a feature; it is the primitive that powers our entire ecosystem flywheel: - Incentive Mechanism: Democratizing contribution and rewarding the trainers and developers. - Assetization: Turning proprietary data and refined models into liquid, ownable assets. - Verifiable Workflow: We are opening the "Black Box" of AI. By bringing total transparency to the Task Generation → Data Collection → Model Training pipeline, we ensure every byte of intelligence is verifiable, traceable, and secure. 2026 Strategic Deliverables This year, we are committed to delivering three foundational pillars: - The World's Largest Training Dataset for Robots: A robot training set—diverse, high-quality interaction data at an unprecedented scale. - A Robotics Foundation Model: A universal robotic brain trained on our pure simulation and synthetic data, capable of robust cross-embodiment transfer and open-world adaptability. - Evolvable Robot Hardware: Robots deployed with Axis models that autonomously evolve through continuous interaction, turning every deployment into a self-improving node within our RGI network. The Ultimate Vision We are building more than models; we are architecting the Distributed Machine Economy. A future where every dataset, model, and robotic embodiment is a verifiable asset in a global, autonomous network. Thank you for building the future of intelligence with us✌️📷

Axis Robotics

28,096 views • 9 months ago

Is Google taking initial steps to enhance Street View? For some reason, Street View seems stuck in technology that feels outdated. I wonder if we'll see such improvements on the product side. Also, note how much better it performs in all aspects compared to Zip-NeRF in their presented material. It offers more details and fewer artifacts. Great work! "LODGE: Level-of-Detail Large-Scale Gaussian Splatting with Efficient Rendering" Contributions: • We propose a novel LOD representation for 3DGS which, unlike previous methods [27, 28, 17], does not recompute the list of used Gaussians at each frame. This allows for acceleration and compaction, enabling the rendering of large-scale scenes even on mobile devices. • We design a strategy to automatically select optimal hyperparameters for splitting LODs, whereas most other methods require manual tuning of hyperparameters for each 3D scene. • To further accelerate rendering, we split the scene into chunks and pre-compute sets of active Gaussians per chunk. • Finally, we introduce a novel opacity interpolation scheme to produce visually pleasing rendering and eliminate artifacts when transitioning between chunks.

Is Google taking initial steps to enhance Street View? For some reason, Street View seems stuck in technology that feels outdated. I wonder if we'll see such improvements on the product side. Also, note how much better it performs in all aspects compared to Zip-NeRF in their presented material. It offers more details and fewer artifacts. Great work! "LODGE: Level-of-Detail Large-Scale Gaussian Splatting with Efficient Rendering" Contributions: • We propose a novel LOD representation for 3DGS which, unlike previous methods [27, 28, 17], does not recompute the list of used Gaussians at each frame. This allows for acceleration and compaction, enabling the rendering of large-scale scenes even on mobile devices. • We design a strategy to automatically select optimal hyperparameters for splitting LODs, whereas most other methods require manual tuning of hyperparameters for each 3D scene. • To further accelerate rendering, we split the scene into chunks and pre-compute sets of active Gaussians per chunk. • Finally, we introduce a novel opacity interpolation scheme to produce visually pleasing rendering and eliminate artifacts when transitioning between chunks.

MrNeRF

62,564 views • 1 year ago