正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Amazon’s training humanoids to carry boxes. I’m racking my brain over why they’d do that. ResMimic enables precise, expressive humanoid loco-manipulation, bridging gaps in general motion tracking (GMT) policies, which lack object awareness. Amazon FAR led the work with a multi-institutional team:

The Humanoid Hub

96,237 subscribers

94,752 次观看 • 11 个月前 •via X (Twitter)

科学技术新闻政治

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Amazon is training humanoids to move boxes. Makes sense! OmniRetarget is a data generation engine that enables complex loco-manipulation for humanoids. It uses offline retargeting from human MoCap datasets and augments data from single demos to produce 8 hours of trajectories that train RL policies in simulation, transferred zero-shot to real robots. Amazon FAR is the primary affiliation for the paper:

Amazon is training humanoids to move boxes. Makes sense! OmniRetarget is a data generation engine that enables complex loco-manipulation for humanoids. It uses offline retargeting from human MoCap datasets and augments data from single demos to produce 8 hours of trajectories that train RL policies in simulation, transferred zero-shot to real robots. Amazon FAR is the primary affiliation for the paper:

The Humanoid Hub

106,447 次观看 • 11 个月前

How do you give a humanoid the general motion capability? Not just single motions, but all motion? Introducing SONIC, our new work on supersizing motion tracking for natural humanoid control. We argue that motion tracking is the scalable foundation task for humanoids. So we "supersized" it: 9k+ GPU hours and 100M+ motion frames. But tracking alone is not enough; we show how to make a useful control system out of it: - Universal Kinematic Planner: Enables game-like gamepad control and high-level teleoperation, just like controlling a character in a game. - VR Full-Body Teleop: Direct, real-time whole-body control by a human wearing a VR headset. - VR Keypoint Teleop: Control the upper body (hands/head) while our planner handles robust locomotion automatically. - VLA Integration: We connect this motion tracker to autonomous Visual-Language-Action (VLA) models for autonomous task execution! We use a Universal Token Space to UNIFY this command space, turning our robust tracker into a general-purpose, programmable humanoid brain. This is the generalist "System 1" for humanoids. 🚀 Project: #Humanoids #Robotics #AI #FoundationModels #NVIDIAResearch 🧠🔥

How do you give a humanoid the general motion capability? Not just single motions, but all motion? Introducing SONIC, our new work on supersizing motion tracking for natural humanoid control. We argue that motion tracking is the scalable foundation task for humanoids. So we "supersized" it: 9k+ GPU hours and 100M+ motion frames. But tracking alone is not enough; we show how to make a useful control system out of it: - Universal Kinematic Planner: Enables game-like gamepad control and high-level teleoperation, just like controlling a character in a game. - VR Full-Body Teleop: Direct, real-time whole-body control by a human wearing a VR headset. - VR Keypoint Teleop: Control the upper body (hands/head) while our planner handles robust locomotion automatically. - VLA Integration: We connect this motion tracker to autonomous Visual-Language-Action (VLA) models for autonomous task execution! We use a Universal Token Space to UNIFY this command space, turning our robust tracker into a general-purpose, programmable humanoid brain. This is the generalist "System 1" for humanoids. 🚀 Project: #Humanoids #Robotics #AI #FoundationModels #NVIDIAResearch 🧠🔥

Zhengyi “Zen” Luo

62,703 次观看 • 10 个月前

The power of generative models — now embodied in humanoids. Announcing DreamControl –– After a year-long research effort at General Robotics — we present a scalable framework for whole-body humanoid control that fuses diffusion priors with reinforcement learning to unlock real-world scene interaction. Diffusion + RL → natural whole-body skills on real robots. DreamControl enables humanoids to move beyond locomotion demos → performing natural, human-like skills such as –– Picking & lifting objects, Opening drawers & doors, Precise punching, kicking, and jumping, Bimanual manipulation tasks Our key innovation: a diffusion prior over human motion that guides RL, eliminating the need for massive teleoperation datasets, and producing motions that look human while transferring to real hardware. Trained purely in simulation, deployed on the Unitree G1 humanoid, DreamControl policies run in real time, bridging sim-to-real with unprecedented naturalness. We leverage a novel hybrid edge + cloud infrastructure that runs RL-trained policies on the edge backed by powerful AI models running in the cloud This is the next step in General Robotics’ journey toward general-purpose humanoid assistants that interact, adapt, and assist autonomously. Paper: Blog: 1/n

The power of generative models — now embodied in humanoids. Announcing DreamControl –– After a year-long research effort at General Robotics — we present a scalable framework for whole-body humanoid control that fuses diffusion priors with reinforcement learning to unlock real-world scene interaction. Diffusion + RL → natural whole-body skills on real robots. DreamControl enables humanoids to move beyond locomotion demos → performing natural, human-like skills such as –– Picking & lifting objects, Opening drawers & doors, Precise punching, kicking, and jumping, Bimanual manipulation tasks Our key innovation: a diffusion prior over human motion that guides RL, eliminating the need for massive teleoperation datasets, and producing motions that look human while transferring to real hardware. Trained purely in simulation, deployed on the Unitree G1 humanoid, DreamControl policies run in real time, bridging sim-to-real with unprecedented naturalness. We leverage a novel hybrid edge + cloud infrastructure that runs RL-trained policies on the edge backed by powerful AI models running in the cloud This is the next step in General Robotics’ journey toward general-purpose humanoid assistants that interact, adapt, and assist autonomously. Paper: Blog: 1/n

Ashish Kapoor

118,467 次观看 • 1 年前

In their latest video, Boston Dynamics’s AI team explains how they make the Atlas humanoid perceive and interact with the world. Atlas uses an agile perception system to understand both the shape and context of objects in complex environments. Atlas combines 2D and 3D awareness, keypoint-based localization, and an object pose tracking system that fuses vision, kinematics, and object knowledge to handle occlusion and uncertainty. Accurate calibration ensures precise hand–eye coordination for reliable manipulation. The team is now working toward a unified model that merges perception and control – pushing beyond spatial AI toward physical intelligence.

In their latest video, Boston Dynamics’s AI team explains how they make the Atlas humanoid perceive and interact with the world. Atlas uses an agile perception system to understand both the shape and context of objects in complex environments. Atlas combines 2D and 3D awareness, keypoint-based localization, and an object pose tracking system that fuses vision, kinematics, and object knowledge to handle occlusion and uncertainty. Accurate calibration ensures precise hand–eye coordination for reliable manipulation. The team is now working toward a unified model that merges perception and control – pushing beyond spatial AI toward physical intelligence.

The Humanoid Hub

50,160 次观看 • 1 年前

Humanoids finally move like humans… and can do more than copy. [Details + demos in thread 👇] A new framework, BeyondMimic, shows how to learn naturalistic whole-body control from human motion. But then goes further by composing those skills into versatile, zero-shot policies. Two key pieces make it work: •Motion tracking pipeline: robustly turns human mocap into dynamic robot motions (jump spins, sprinting, cartwheels) on real hardware •Guided diffusion policy: distills motion primitives and enables zero-shot downstream tasks with simple cost functions The result: ✅ State-of-the-art motion quality from mocap tracking ✅ Versatile task execution; waypoint navigation, joystick teleop, obstacle avoidance ✅ Stable sim-to-real transfer for diverse humanoid skills Thank you for sharing, Qiayuan Liao! 📍 Paper: Video:

Humanoids finally move like humans… and can do more than copy. [Details + demos in thread 👇] A new framework, BeyondMimic, shows how to learn naturalistic whole-body control from human motion. But then goes further by composing those skills into versatile, zero-shot policies. Two key pieces make it work: •Motion tracking pipeline: robustly turns human mocap into dynamic robot motions (jump spins, sprinting, cartwheels) on real hardware •Guided diffusion policy: distills motion primitives and enables zero-shot downstream tasks with simple cost functions The result: ✅ State-of-the-art motion quality from mocap tracking ✅ Versatile task execution; waypoint navigation, joystick teleop, obstacle avoidance ✅ Stable sim-to-real transfer for diverse humanoid skills Thank you for sharing, Qiayuan Liao! 📍 Paper: Video:

Ilir Aliu - eu/acc

72,522 次观看 • 1 年前

SceNeRFlow: Time-Consistent Reconstruction of General Dynamic Scenes abs: paper page: Existing methods for the 4D reconstruction of general, non-rigidly deforming objects focus on novel-view synthesis and neglect correspondences. However, time consistency enables advanced downstream tasks like 3D editing, motion analysis, or virtual-asset creation. We propose SceNeRFlow to reconstruct a general, non-rigid scene in a time-consistent manner. Our dynamic-NeRF method takes multi-view RGB videos and background images from static cameras with known camera parameters as input. It then reconstructs the deformations of an estimated canonical model of the geometry and appearance in an online fashion. Since this canonical model is time-invariant, we obtain correspondences even for long-term, long-range motions. We employ neural scene representations to parametrize the components of our method. Like prior dynamic-NeRF methods, we use a backwards deformation model. We find non-trivial adaptations of this model necessary to handle larger motions: We decompose the deformations into a strongly regularized coarse component and a weakly regularized fine component, where the coarse component also extends the deformation field into the space surrounding the object, which enables tracking over time. We show experimentally that, unlike prior work that only handles small motion, our method enables the reconstruction of studio-scale motions.

SceNeRFlow: Time-Consistent Reconstruction of General Dynamic Scenes abs: paper page: Existing methods for the 4D reconstruction of general, non-rigidly deforming objects focus on novel-view synthesis and neglect correspondences. However, time consistency enables advanced downstream tasks like 3D editing, motion analysis, or virtual-asset creation. We propose SceNeRFlow to reconstruct a general, non-rigid scene in a time-consistent manner. Our dynamic-NeRF method takes multi-view RGB videos and background images from static cameras with known camera parameters as input. It then reconstructs the deformations of an estimated canonical model of the geometry and appearance in an online fashion. Since this canonical model is time-invariant, we obtain correspondences even for long-term, long-range motions. We employ neural scene representations to parametrize the components of our method. Like prior dynamic-NeRF methods, we use a backwards deformation model. We find non-trivial adaptations of this model necessary to handle larger motions: We decompose the deformations into a strongly regularized coarse component and a weakly regularized fine component, where the coarse component also extends the deformation field into the space surrounding the object, which enables tracking over time. We show experimentally that, unlike prior work that only handles small motion, our method enables the reconstruction of studio-scale motions.

AK

76,380 次观看 • 3 年前

This is not fair Andy Jassy cc Jeff Bezos Amazon is charging my company money to answer customer questions and then directing our customers to inferior competitor products who have paid Amazon. We never signed up for Rufus ads. The customer is already on the Amazon page for Brain Flakes. We may have even already paid for an ad for them to get there. And then Amazon shows them an ad designed to answer questions that are already answered by us directly on the page they are on? And then Amazon sends them to another “brand” which is advertising on our product page. Look, I get it. Amazon is a big company with lots of people, lots of teams, and lots of AI, but you can’t keep doing this. You have to rethink your approach. The more you squeeze sellers, the higher prices will go on Amazon, because we don’t have the margins that you do (or the access to capital). This combined with Amazon’s market share and pricing policies mean that when you force us to raise prices by raising our costs, pricing across the entire ecommerce ecosystem rise. In other words, all Americans pay more when shopping online. You need to rethink these policies and the endless backdoor price increases. Thank you

molson 🧠⚙️

43,116 次观看 • 5 个月前

🎙️ Excited to introduce one of my favorite projects from the past year: TeleDexter, from the BIGAI dexterity team. It’s a stable, human-level dexterous teleoperation system and a suite of autonomous policies trained with it. Pen spinning, complex in-hand reorientation, and long-horizon tool use—once seen as the holy grail of manipulation—are now unlocked. 🧵👇 The hardware is already here; we have some incredible high-DoF robotic hands. The bottleneck? The controller. Most current systems are stuck in "quasi-static" grasping mode. Meanwhile, dynamic in-hand dexterity has remained severely limited. 🧠 To unlock the massive capabilities of human-like hands, we need to build an excellent "cerebellum" for dexterous hands. TeleDexter solves this with a novel co-tracking approach: it simultaneously tracks both human hand kinematics and object states, beautifully bridging the gap between human intent and robotic control. In order to train a better co-tracking policy that works robustly in the real world, we designed : (1) a hybrid reward design that combines consecutive goal reaching and dense tracking, (2) an action masking strategy during training that enhances sim2real performance, (3) a dexterous curriculum for learning the long-horizon interactions. Each design is inspired by numerous trials and countless real-world experiments. We’ve synthesized all the system details, engineering challenges, and core insights into our latest post. If you're interested in the future of dexterous manipulation, grab a coffee and check it out (9-min read): If you have more time, check out the paper:

🎙️ Excited to introduce one of my favorite projects from the past year: TeleDexter, from the BIGAI dexterity team. It’s a stable, human-level dexterous teleoperation system and a suite of autonomous policies trained with it. Pen spinning, complex in-hand reorientation, and long-horizon tool use—once seen as the holy grail of manipulation—are now unlocked. 🧵👇 The hardware is already here; we have some incredible high-DoF robotic hands. The bottleneck? The controller. Most current systems are stuck in "quasi-static" grasping mode. Meanwhile, dynamic in-hand dexterity has remained severely limited. 🧠 To unlock the massive capabilities of human-like hands, we need to build an excellent "cerebellum" for dexterous hands. TeleDexter solves this with a novel co-tracking approach: it simultaneously tracks both human hand kinematics and object states, beautifully bridging the gap between human intent and robotic control. In order to train a better co-tracking policy that works robustly in the real world, we designed : (1) a hybrid reward design that combines consecutive goal reaching and dense tracking, (2) an action masking strategy during training that enhances sim2real performance, (3) a dexterous curriculum for learning the long-horizon interactions. Each design is inspired by numerous trials and countless real-world experiments. We’ve synthesized all the system details, engineering challenges, and core insights into our latest post. If you're interested in the future of dexterous manipulation, grab a coffee and check it out (9-min read): If you have more time, check out the paper:

Siyuan Huang

12,602 次观看 • 2 个月前

🦿Xpeng showed a humanoid robot called IRON whose movement looked so human that the team literally cut it open on stage to prove it is a machine. IRON uses a bionic body with a flexible spine, synthetic muscles, and soft skin so joints and torso can twist smoothly like a person. The system has 82 degrees of freedom in total with 22 in each hand for fine finger control. Compute runs on 3 custom AI chips rated at 2,250 TOPS (Tera Operations Per Second), which is far above typical laptop neural accelerators, so it can handle vision and motion planning on the robot. The AI stack focuses on turning camera input directly into body movement without routing through text, which reduces lag and makes the gait look natural. Xpeng staged the cut-open demo at AI Day in Guangzhou this week, addressing rumors that a performer was inside by exposing internal actuators, wiring, and cooling. Company materials also mention a large physical-world model and a multi-brain control setup for dialogue, perception, and locomotion, hinting at a path from stage demos to service work. Production is targeted for 2026, so near-term tasks will be limited, but the hardware shows a serious step toward human-scale manipulation.

🦿Xpeng showed a humanoid robot called IRON whose movement looked so human that the team literally cut it open on stage to prove it is a machine. IRON uses a bionic body with a flexible spine, synthetic muscles, and soft skin so joints and torso can twist smoothly like a person. The system has 82 degrees of freedom in total with 22 in each hand for fine finger control. Compute runs on 3 custom AI chips rated at 2,250 TOPS (Tera Operations Per Second), which is far above typical laptop neural accelerators, so it can handle vision and motion planning on the robot. The AI stack focuses on turning camera input directly into body movement without routing through text, which reduces lag and makes the gait look natural. Xpeng staged the cut-open demo at AI Day in Guangzhou this week, addressing rumors that a performer was inside by exposing internal actuators, wiring, and cooling. Company materials also mention a large physical-world model and a multi-brain control setup for dialogue, perception, and locomotion, hinting at a path from stage demos to service work. Production is targeted for 2026, so near-term tasks will be limited, but the hardware shows a serious step toward human-scale manipulation.

Rohan Paul

3,802,543 次观看 • 10 个月前

This work makes a humanoid robot do simple parkour moves by looking with a depth camera and choosing the right move on the fly. The big deal is that it turns lots of small human moves into long, real-time robot behavior, without hand-coding every transition or retraining for each new course. A humanoid robot is usually good at steady walking, but it often fails when it has to do fast moves like jumping up, vaulting, or rolling, and then keep going to the next obstacle. The hard part is that you cannot easily collect training data for every possible obstacle shape, distance, and mistake, so robots end up learning a few moves that only work in a narrow setup. This work starts from short clips of real human parkour moves, like stepping over, vaulting, climbing, and rolling. It uses motion matching, which is basically a smart “pick the next clip that fits best right now” search, to stitch those short clips into a long, smooth plan that looks like a human doing a whole course. Then it trains a controller with reinforcement learning (RL), which means the robot learns by trial and error to copy that plan while staying balanced and not falling. After training separate expert controllers for different moves, it compresses them into 1 controller that uses only onboard depth sensing and a simple “go this fast in this direction” command. In real tests on a Unitree G1 humanoid, it can clear multiple obstacles in a row, adapt when obstacles get moved, and climb a wall up to 1.25m.

This work makes a humanoid robot do simple parkour moves by looking with a depth camera and choosing the right move on the fly. The big deal is that it turns lots of small human moves into long, real-time robot behavior, without hand-coding every transition or retraining for each new course. A humanoid robot is usually good at steady walking, but it often fails when it has to do fast moves like jumping up, vaulting, or rolling, and then keep going to the next obstacle. The hard part is that you cannot easily collect training data for every possible obstacle shape, distance, and mistake, so robots end up learning a few moves that only work in a narrow setup. This work starts from short clips of real human parkour moves, like stepping over, vaulting, climbing, and rolling. It uses motion matching, which is basically a smart “pick the next clip that fits best right now” search, to stitch those short clips into a long, smooth plan that looks like a human doing a whole course. Then it trains a controller with reinforcement learning (RL), which means the robot learns by trial and error to copy that plan while staying balanced and not falling. After training separate expert controllers for different moves, it compresses them into 1 controller that uses only onboard depth sensing and a simple “go this fast in this direction” command. In real tests on a Unitree G1 humanoid, it can clear multiple obstacles in a row, adapt when obstacles get moved, and climb a wall up to 1.25m.

Rohan Paul

37,121 次观看 • 7 个月前

Today, we publicly released RoboCasa365, a large-scale simulation benchmark for training and systematically evaluating generalist robot models. Built upon our original RoboCasa framework, it offers: • 2,500 realistic kitchen environments; • 365 everyday tasks (basic skills + long-horizon mobile manipulation); • Over 3,200 objects with many articulated fixtures/appliances. All are designed for fully controlled, reproducible benchmarking of robotic policies. Progress in robotic foundation models is real. But it’s still hard to answer basic questions like: How close are we to general-purpose autonomy? What factors drive generalization? What are the model/data scaling curves like? Real-world eval is slow and noisy, and existing sims (like LIBERO, which we built 3 years ago) often lack sufficient task and scene diversity. This benchmark comes with 2,200+ hours of demonstrations and 500K+ trajectories to support studies of multi-task training, pretraining, and continual learning at scale. Check it out at

Today, we publicly released RoboCasa365, a large-scale simulation benchmark for training and systematically evaluating generalist robot models. Built upon our original RoboCasa framework, it offers: • 2,500 realistic kitchen environments; • 365 everyday tasks (basic skills + long-horizon mobile manipulation); • Over 3,200 objects with many articulated fixtures/appliances. All are designed for fully controlled, reproducible benchmarking of robotic policies. Progress in robotic foundation models is real. But it’s still hard to answer basic questions like: How close are we to general-purpose autonomy? What factors drive generalization? What are the model/data scaling curves like? Real-world eval is slow and noisy, and existing sims (like LIBERO, which we built 3 years ago) often lack sufficient task and scene diversity. This benchmark comes with 2,200+ hours of demonstrations and 500K+ trajectories to support studies of multi-task training, pretraining, and continual learning at scale. Check it out at

Yuke Zhu

24,394 次观看 • 6 个月前

Jarvis Landry says after getting paid $75,000,000 and joining the Browns, he deliberately kept training like he was “dead broke” to set a standard “I played around a lot of talented players that also had a different mindset, bro. They just thought different. When you’re around like-minded people, or when you’re around people that think like that, either you’ve got to catch up or you’re going to get kicked off the ship.” “That was something that I developed, and that’s something that I still had in Miami. Miami had a little bit more of a culture. We had older players like Cameron Wake. I say his name first off the top of my head because that’s one of the leaders that was on the team. He spoke in the huddles, he spoke in team meetings and things like that.” “Then when I got to Cleveland, it was like none of that was around, literally. Everybody was looking around like, ‘Who’s going to say anything? Is anybody going to say anything?’ I didn’t say that I saw an opportunity there to go in and give a speech or whatever.” “That led from us training through the offseason. Once I got signed there, I started flying guys here, just trying to do things together, see who I’m going to work with, learning about these guys, seeing how hard they work. Do they think like I think? Showing them, ‘Yeah, I just got paid, but I’m still out here running routes in offseason training.’ Somebody working like I’m dead broke.” “So when we got to training camp, we practiced one day and guys are like, ‘my hamstring sore.’ I’m like, ‘So what have you been doing this whole time, fam?’ The next day we get the script, you see the numbers and a couple guys aren’t practicing. I’m like, ‘So why?’ ‘His hamstring is sore,’ or his quad, or his feet hurt. I’m like, ‘You could take practice off for being sore?’” “I went in the training room and the trainer explained to me about a couple guys that had some chronic issues, mostly linemen. The other shit, I’m like, So what?’ He’s like, ‘We don’t want to, it’s too early to get guys hurt.’ That kind of ticked me off because my hamstrings hurt too. My hip flexors hurt too. My feet hurt too.” “That was kind of what led to me giving the ‘weakness is contagious speech’. Bro, I can’t be great that way. If every time I feel resistance or pain, if I always just pull back from it to try to feel better, I’m not going to get past what I need to get past to try to be great.”

Jarvis Landry says after getting paid $75,000,000 and joining the Browns, he deliberately kept training like he was “dead broke” to set a standard “I played around a lot of talented players that also had a different mindset, bro. They just thought different. When you’re around like-minded people, or when you’re around people that think like that, either you’ve got to catch up or you’re going to get kicked off the ship.” “That was something that I developed, and that’s something that I still had in Miami. Miami had a little bit more of a culture. We had older players like Cameron Wake. I say his name first off the top of my head because that’s one of the leaders that was on the team. He spoke in the huddles, he spoke in team meetings and things like that.” “Then when I got to Cleveland, it was like none of that was around, literally. Everybody was looking around like, ‘Who’s going to say anything? Is anybody going to say anything?’ I didn’t say that I saw an opportunity there to go in and give a speech or whatever.” “That led from us training through the offseason. Once I got signed there, I started flying guys here, just trying to do things together, see who I’m going to work with, learning about these guys, seeing how hard they work. Do they think like I think? Showing them, ‘Yeah, I just got paid, but I’m still out here running routes in offseason training.’ Somebody working like I’m dead broke.” “So when we got to training camp, we practiced one day and guys are like, ‘my hamstring sore.’ I’m like, ‘So what have you been doing this whole time, fam?’ The next day we get the script, you see the numbers and a couple guys aren’t practicing. I’m like, ‘So why?’ ‘His hamstring is sore,’ or his quad, or his feet hurt. I’m like, ‘You could take practice off for being sore?’” “I went in the training room and the trainer explained to me about a couple guys that had some chronic issues, mostly linemen. The other shit, I’m like, So what?’ He’s like, ‘We don’t want to, it’s too early to get guys hurt.’ That kind of ticked me off because my hamstrings hurt too. My hip flexors hurt too. My feet hurt too.” “That was kind of what led to me giving the ‘weakness is contagious speech’. Bro, I can’t be great that way. If every time I feel resistance or pain, if I always just pull back from it to try to feel better, I’m not going to get past what I need to get past to try to be great.”

Josh Chambers

520,724 次观看 • 1 个月前

We are excited to share our latest work, "Superhuman Safe and Agile Racing through Multi-Agent Reinforcement Learning," done in collaboration with Google DeepMind . Autonomous drones have reached superhuman speed in isolation, but what happens when multiple agents share the same airspace? Paper: Website: Video: Using league-based self-play, we train #ReinforcementLearning agents that race against a diverse, evolving population of opponents. Through this competitive training, sophisticated behaviors emerge without explicit programming: strategic overtaking, proactive collision avoidance, and even awareness of aerodynamic downwash from nearby drones. In real-world multi-player races at speeds exceeding 80kph (50 mph) and accelerations up to 7g, our agents outperform a five-time Swiss national drone racing champion while reducing collision rates by 50% compared to single-agent baselines. Crucially, training against diverse artificial opponents enables zero-shot generalization to human pilots, achieving over 90% race completion in mixed human-AI races with up to four competitors. A key insight: human pilots adopt riskier strategies when trailing, leading to more crashes under competitive pressure. Our learned policies, by contrast, maintain consistent safety margins regardless of race standing, a property essential for deploying autonomous systems alongside humans. Also, the multi-agent self-play policies are more robust than those trained independently, suggesting that training in competitive environments is not only key to winning races but also to learning safer, more reliable autonomy for real-world multi-robot systems. Kudos to Ismail Geles, Leonard Bauersfeld, Markus Wulfmeier! Ismail Geles Leonard Bauersfeld Markus Wulfmeier European Research Council (ERC) UZH IfI University of Zurich UZH Science UZH Space Hub Swiss Robotics NCCR Robotics

We are excited to share our latest work, "Superhuman Safe and Agile Racing through Multi-Agent Reinforcement Learning," done in collaboration with Google DeepMind . Autonomous drones have reached superhuman speed in isolation, but what happens when multiple agents share the same airspace? Paper: Website: Video: Using league-based self-play, we train #ReinforcementLearning agents that race against a diverse, evolving population of opponents. Through this competitive training, sophisticated behaviors emerge without explicit programming: strategic overtaking, proactive collision avoidance, and even awareness of aerodynamic downwash from nearby drones. In real-world multi-player races at speeds exceeding 80kph (50 mph) and accelerations up to 7g, our agents outperform a five-time Swiss national drone racing champion while reducing collision rates by 50% compared to single-agent baselines. Crucially, training against diverse artificial opponents enables zero-shot generalization to human pilots, achieving over 90% race completion in mixed human-AI races with up to four competitors. A key insight: human pilots adopt riskier strategies when trailing, leading to more crashes under competitive pressure. Our learned policies, by contrast, maintain consistent safety margins regardless of race standing, a property essential for deploying autonomous systems alongside humans. Also, the multi-agent self-play policies are more robust than those trained independently, suggesting that training in competitive environments is not only key to winning races but also to learning safer, more reliable autonomy for real-world multi-robot systems. Kudos to Ismail Geles, Leonard Bauersfeld, Markus Wulfmeier! Ismail Geles Leonard Bauersfeld Markus Wulfmeier European Research Council (ERC) UZH IfI University of Zurich UZH Science UZH Space Hub Swiss Robotics NCCR Robotics

Davide Scaramuzza

14,742 次观看 • 4 个月前

Big Mac passed yesterday. Decided not much point sitting around. There’s dogs to save… Saving street dogs is my mission but to do that takes a lot of money and a lot of awareness. I wanted to do something to really push things to another level… 🏃‍♂️ I will run a marathon on the 15th November. I have only started training today with 30 days to go. 🏙️ The race is in Bangkok at night. The most I have even run Iy life is 20km 🥘 I have been eating really badly lately and am very unfit. I’m going to share the journey over the next 30 days in what could be one of the stupidest ideas I’ve ever had. I have 3 main reasons to do it… 1. I was to raise awareness for Happy Doggo as a charity. (A follow there helps) 2. I want to try raise money for a mobile sterilizing unit with 10 staff and 2 trucks treating 5000 dogs a year. More on that soon. 3. My mental health hasn’t been amazing lately with so many losses so I wanted a good physical challenge to clear my head. I expect people to say it’s impossible and try talk me out of it like the team here all have! The chances are I’ll probably fail but life is short and I’ve a lot of dogs to help. I got a 5km in today to start which was tough but we are off and running. I’ll be sharing this series over on Happy Doggo daily 🤞🤞🤞

Big Mac passed yesterday. Decided not much point sitting around. There’s dogs to save… Saving street dogs is my mission but to do that takes a lot of money and a lot of awareness. I wanted to do something to really push things to another level… 🏃‍♂️ I will run a marathon on the 15th November. I have only started training today with 30 days to go. 🏙️ The race is in Bangkok at night. The most I have even run Iy life is 20km 🥘 I have been eating really badly lately and am very unfit. I’m going to share the journey over the next 30 days in what could be one of the stupidest ideas I’ve ever had. I have 3 main reasons to do it… 1. I was to raise awareness for Happy Doggo as a charity. (A follow there helps) 2. I want to try raise money for a mobile sterilizing unit with 10 staff and 2 trucks treating 5000 dogs a year. More on that soon. 3. My mental health hasn’t been amazing lately with so many losses so I wanted a good physical challenge to clear my head. I expect people to say it’s impossible and try talk me out of it like the team here all have! The chances are I’ll probably fail but life is short and I’ve a lot of dogs to help. I got a 5km in today to start which was tough but we are off and running. I’ll be sharing this series over on Happy Doggo daily 🤞🤞🤞

Niall Harbison

183,861 次观看 • 11 个月前

I don’t know if we live in a Matrix, but I know for sure that robots will spend most of their lives in simulation. Let machines train machines. I’m excited to introduce DexMimicGen, a massive-scale synthetic data generator that enables a humanoid robot to learn complex skills from only a handful of human demonstrations. Yes, as few as 5! DexMimicGen addresses the biggest pain point in robotics: where do we get data? Unlike with LLMs, where vast amounts of texts are readily available, you cannot simply download motor control signals from the internet. So researchers teleoperate the robots to collect motion data via XR headsets. They have to repeat the same skill over and over and over again, because neural nets are data hungry. This is a very slow and uncomfortable process. At NVIDIA, we believe the majority of high-quality tokens for robot foundation models will come from simulation. What DexMimicGen does is to trade GPU compute time for human time. It takes one motion trajectory from human, and multiplies into 1000s of new trajectories. A robot brain trained on this augmented dataset will generalize far better in the real world. Think of DexMimicGen as a learning signal amplifier. It maps a small dataset to a large (de facto infinite) dataset, using physics simulation in the loop. In this way, we free humans from babysitting the bots all day. The future of robot data is generative. The future of the entire robot learning pipeline will also be generative. 🧵

I don’t know if we live in a Matrix, but I know for sure that robots will spend most of their lives in simulation. Let machines train machines. I’m excited to introduce DexMimicGen, a massive-scale synthetic data generator that enables a humanoid robot to learn complex skills from only a handful of human demonstrations. Yes, as few as 5! DexMimicGen addresses the biggest pain point in robotics: where do we get data? Unlike with LLMs, where vast amounts of texts are readily available, you cannot simply download motor control signals from the internet. So researchers teleoperate the robots to collect motion data via XR headsets. They have to repeat the same skill over and over and over again, because neural nets are data hungry. This is a very slow and uncomfortable process. At NVIDIA, we believe the majority of high-quality tokens for robot foundation models will come from simulation. What DexMimicGen does is to trade GPU compute time for human time. It takes one motion trajectory from human, and multiplies into 1000s of new trajectories. A robot brain trained on this augmented dataset will generalize far better in the real world. Think of DexMimicGen as a learning signal amplifier. It maps a small dataset to a large (de facto infinite) dataset, using physics simulation in the loop. In this way, we free humans from babysitting the bots all day. The future of robot data is generative. The future of the entire robot learning pipeline will also be generative. 🧵

Jim Fan

165,246 次观看 • 1 年前

This is how ALOHA's "teleoperation" system works - a fancy word for "remote control". Training robots will be more and more like playing games in the physical world. A human operates a "joystick++" to perform tasks and collect data, or intervene if there's any safety concern. There's actually a learning curve to master the controller, much like practicing gaming skills. Teleoperation can be done in many different ways. ALOHA is an impressive custom-built system with very low cost. Here're a few alternatives: (1) Motion Capture (MoCap): apply the MoCap systems used for Hollywood movies to capture the fine-grained motions of hand joints. There would be no "embodiment gap" if the robot hand has 5 fingers. For instance, a demonstrator can wear a CyberGlove ( and manipulate the objects. CyberGlove will capture the motion signals & haptic feedback in real-time, which can be re-targeted onto the humanoid. (2) Wearing gloves & markers can be clumsy. An alternative way to do MoCap is through computer vision. DexPilot from NVIDIA enables marker-less and glove-free data collection. The human operator simply uses their bare hands to perform the tasks. 4 Intel RealSense depth cameras and 2 NVIDIA Titan XP GPUs (yeah, 2019 work) translate the pixels to precise motion signals for robot learning. (3) VR Headset: turn the training room into a VR game and "role play" the robot. This has the advantage of scalable remote data collection - annotators from around the world can contribute without coming onsite. VR demonstration technique appeared in research projects like the iGibson home robot simulator, an initiative that I participated in at Stanford: Behind-the-scene video by Litian Liang

This is how ALOHA's "teleoperation" system works - a fancy word for "remote control". Training robots will be more and more like playing games in the physical world. A human operates a "joystick++" to perform tasks and collect data, or intervene if there's any safety concern. There's actually a learning curve to master the controller, much like practicing gaming skills. Teleoperation can be done in many different ways. ALOHA is an impressive custom-built system with very low cost. Here're a few alternatives: (1) Motion Capture (MoCap): apply the MoCap systems used for Hollywood movies to capture the fine-grained motions of hand joints. There would be no "embodiment gap" if the robot hand has 5 fingers. For instance, a demonstrator can wear a CyberGlove ( and manipulate the objects. CyberGlove will capture the motion signals & haptic feedback in real-time, which can be re-targeted onto the humanoid. (2) Wearing gloves & markers can be clumsy. An alternative way to do MoCap is through computer vision. DexPilot from NVIDIA enables marker-less and glove-free data collection. The human operator simply uses their bare hands to perform the tasks. 4 Intel RealSense depth cameras and 2 NVIDIA Titan XP GPUs (yeah, 2019 work) translate the pixels to precise motion signals for robot learning. (3) VR Headset: turn the training room into a VR game and "role play" the robot. This has the advantage of scalable remote data collection - annotators from around the world can contribute without coming onsite. VR demonstration technique appeared in research projects like the iGibson home robot simulator, an initiative that I participated in at Stanford: Behind-the-scene video by Litian Liang

Jim Fan

124,783 次观看 • 2 年前