Загрузка видео...

Не удалось загрузить видео

Возникла проблема при загрузке этого видео. Это может быть связано с временными проблемами сети или видео может быть недоступно.

На главную

Interested in making your bipedal robots to be athletes? We summarized our RL work to create robust & adaptive controllers for general bipedal skills. 400m-dash, running over terrains/against perturbations, targeted jumping, compliant walking, not a problem for bipeds now.🧵👇

Zhongyu Li

1,499 subscribers

86,743 просмотров • 2 лет назад •via X (Twitter)

Образование Спорт Наука и технологии

Anya Rossi• Live Now

Private livecam show

Комментарии: 10

Фото профиля Zhongyu Li

Zhongyu Li2 лет назад

We found that RL can bring robustness and adaptivity for dynamic & real-time bipedal control, but only with a proper formulation. Using a long history of robot I/O, complementing it with a short history, training it end-end, and adding task randomization, are the recipes.

Фото профиля Zhongyu Li

Zhongyu Li2 лет назад

We conducted extensive benchmark and ablation studies in both sim and real to support this. Details are in the paper: Summary video: This work cannot be done without @xbpeng4 @GlenBerseth @svlevine @pabbeel & Koushil

Фото профиля Zhongyu Li

Zhongyu Li2 лет назад

There are more cool videos👇 complete 400m dash: more running: more walking:

Фото профиля Guanya Shi

Guanya Shi2 лет назад

Great work! Congratulations!

Фото профиля Glen Berseth

Glen Berseth2 лет назад

We should really apply for a #GuinnessWorldRecord with this work. Fastest 400m sprint with a robot.

Фото профиля echo_name

echo_name2 лет назад

if the little guy learns to run in lane 1 I think it could run sub 2 in the 400m!

Фото профиля Brett Adcock

Brett Adcock2 лет назад

cool

Фото профиля Tony Z. Zhao

Tony Z. Zhao2 лет назад

Congrats Zhongyu!!

Фото профиля Hongtao Gu

Hongtao Gu2 лет назад

This is so cool!

Фото профиля Dr.Among_Us

Dr.Among_Us2 лет назад

🫡🫡

Похожие видео

Agile Continuous Jumping in Discontinuous Terrains discuss: We focus on agile, continuous, and terrain-adaptive jumping of quadrupedal robots in discontinuous terrains such as stairs and stepping stones. Unlike single-step jumping, continuous jumping requires accurately executing highly dynamic motions over long horizons, which is challenging for existing approaches. To accomplish this task, we design a hierarchical learning and control framework, which consists of a learned heightmap predictor for robust terrain perception, a reinforcement-learning-based centroidal-level motion policy for versatile and terrain-adaptive planning, and a low-level model-based leg controller for accurate motion tracking. In addition, we minimize the sim-to-real gap by accurately modeling the hardware characteristics. Our framework enables a Unitree Go1 robot to perform agile and continuous jumps on human-sized stairs and sparse stepping stones, for the first time to the best of our knowledge. In particular, the robot can cross two stair steps in each jump and completes a 3.5m long, 2.8m high, 14-step staircase in 4.5 seconds. Moreover, the same policy outperforms baselines in various other parkour tasks, such as jumping over single horizontal or vertical discontinuities.

Agile Continuous Jumping in Discontinuous Terrains discuss: We focus on agile, continuous, and terrain-adaptive jumping of quadrupedal robots in discontinuous terrains such as stairs and stepping stones. Unlike single-step jumping, continuous jumping requires accurately executing highly dynamic motions over long horizons, which is challenging for existing approaches. To accomplish this task, we design a hierarchical learning and control framework, which consists of a learned heightmap predictor for robust terrain perception, a reinforcement-learning-based centroidal-level motion policy for versatile and terrain-adaptive planning, and a low-level model-based leg controller for accurate motion tracking. In addition, we minimize the sim-to-real gap by accurately modeling the hardware characteristics. Our framework enables a Unitree Go1 robot to perform agile and continuous jumps on human-sized stairs and sparse stepping stones, for the first time to the best of our knowledge. In particular, the robot can cross two stair steps in each jump and completes a 3.5m long, 2.8m high, 14-step staircase in 4.5 seconds. Moreover, the same policy outperforms baselines in various other parkour tasks, such as jumping over single horizontal or vertical discontinuities.

AK

35,794 просмотров • 1 год назад

NAVIAI-I3, a more robust humanoid robot, was launched by the Zhejiang Humanoid Robotics Innovation Center. Standing 1.8 meters tall, it can run like a human and walk on various terrains. This humanoid robot appears to be primarily designed for specialized industrial tasks, such as security patrols in underground parking lots. They previously launched the lightweight bipedal NVAI-2 and a wheeled humanoid robot. The bipedal robot is used in manufacturing operations, while the wheeled humanoid robot is used for unstructured, flexible tasks such as warehouse sorting, supermarket shelving, and office cleaning. We filmed this demonstration at the World Robot Conference; you can see it in the comments section.

NAVIAI-I3, a more robust humanoid robot, was launched by the Zhejiang Humanoid Robotics Innovation Center. Standing 1.8 meters tall, it can run like a human and walk on various terrains. This humanoid robot appears to be primarily designed for specialized industrial tasks, such as security patrols in underground parking lots. They previously launched the lightweight bipedal NVAI-2 and a wheeled humanoid robot. The bipedal robot is used in manufacturing operations, while the wheeled humanoid robot is used for unstructured, flexible tasks such as warehouse sorting, supermarket shelving, and office cleaning. We filmed this demonstration at the World Robot Conference; you can see it in the comments section.

CyberRobo

15,521 просмотров • 8 месяцев назад

You know some white people still see Black people as slaves. They see themselves as modern day slave patrol, i.e. cops.. show us your papers or not you will be in trouble. We’re going to work., going to a place of homes. Business.. out walking., running, everything we do it’s a problem for them. and it’s like a death sentence for us.

You know some white people still see Black people as slaves. They see themselves as modern day slave patrol, i.e. cops.. show us your papers or not you will be in trouble. We’re going to work., going to a place of homes. Business.. out walking., running, everything we do it’s a problem for them. and it’s like a death sentence for us.

Don Salmon

504,531 просмотров • 8 месяцев назад

🧵Seems Cllr Stephen Holyday WILL interrupt & silence a speaker at a public meeting! But only if you ask him to denounce violent threats against cyclists & apologize for his complicity in allowing the idea of running over people on bikes in your way to be applauded 1/4

🧵Seems Cllr Stephen Holyday WILL interrupt & silence a speaker at a public meeting! But only if you ask him to denounce violent threats against cyclists & apologize for his complicity in allowing the idea of running over people on bikes in your way to be applauded 1/4

The Biking Lawyer (Dave Shellnutt)

31,822 просмотров • 2 лет назад

🚨We're hiring a Marketing & Events Associate in SF! You'll work alongside our CMO to engage media, create content, produce podcasts, and run events. Looking for: - 3-6 yrs experience - strong communication skills - crypto background not required Details in comments 👇

🚨We're hiring a Marketing & Events Associate in SF! You'll work alongside our CMO to engage media, create content, produce podcasts, and run events. Looking for: - 3-6 yrs experience - strong communication skills - crypto background not required Details in comments 👇

Framework Ventures

11,913 просмотров • 4 месяцев назад

Researchers at Caltech have built a bipedal robot that combines walking with flying to create a new type of locomotion, making it exceptionally nimble and capable of complex movements. Part walking robot, part flying drone, LEONARDO (short for LEgs ONboARD drOne, or LEO for short) can walk a slackline, hop, and even ride a skateboard. Developed by a team at Caltech's Center for Autonomous Systems and Technologies (CAST), LEO is the first robot that uses multi-joint legs and propeller-based thrusters to achieve a fine degree of control over its balance. Video Credit: Caltech #robots #robotics #engineering #caltech -------------------------------- Stay ahead of the curve! Follow us now on our WhatsApp ( and Telegram ( channels and stay updated about the cutting edge.

Researchers at Caltech have built a bipedal robot that combines walking with flying to create a new type of locomotion, making it exceptionally nimble and capable of complex movements. Part walking robot, part flying drone, LEONARDO (short for LEgs ONboARD drOne, or LEO for short) can walk a slackline, hop, and even ride a skateboard. Developed by a team at Caltech's Center for Autonomous Systems and Technologies (CAST), LEO is the first robot that uses multi-joint legs and propeller-based thrusters to achieve a fine degree of control over its balance. Video Credit: Caltech #robots #robotics #engineering #caltech -------------------------------- Stay ahead of the curve! Follow us now on our WhatsApp ( and Telegram ( channels and stay updated about the cutting edge.

Wevolver

20,044 просмотров • 1 год назад

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

118,133 просмотров • 10 месяцев назад

This is our general call for all emerging artists across Africa 🚂 .. Are you interested in featuring on our playlist cover “Emerging Africa” Instagram teaser interview to share your music to network of A&Rs and music lovers in our community! SEND A DM now 📤 👀

This is our general call for all emerging artists across Africa 🚂 .. Are you interested in featuring on our playlist cover “Emerging Africa” Instagram teaser interview to share your music to network of A&Rs and music lovers in our community! SEND A DM now 📤 👀

A&R DUTY

30,806 просмотров • 2 лет назад

Our job is to lower your bills. Zonal electricity pricing is how we do so - fixing our wasteful, broken system and making it work for you. In the latest Inside Octopus podcast, Greg, and Jason from FTI Consulting get to the bottom of zonal pricing. Save it for today’s commute: 🧵👇

Our job is to lower your bills. Zonal electricity pricing is how we do so - fixing our wasteful, broken system and making it work for you. In the latest Inside Octopus podcast, Greg, and Jason from FTI Consulting get to the bottom of zonal pricing. Save it for today’s commute: 🧵👇

Octopus Energy

19,570 просмотров • 1 год назад

This is the biggest group of Sinn Féin TDs to enter Leinster House in over a century We are here to work for you, we are in your corner and we will never ever give up on our mission to create an equal, a fair and a united Ireland

This is the biggest group of Sinn Féin TDs to enter Leinster House in over a century We are here to work for you, we are in your corner and we will never ever give up on our mission to create an equal, a fair and a united Ireland

Mary Lou McDonald

104,567 просмотров • 1 год назад

Clarendon📍 We nah play! We’re making the necessary repairs to restore your power. While you may not see damage to JPS structures in your area, the problem could be much further up the road. Our teams are in their hometowns working to restore power. Thank you for your patience🤝🏾

Clarendon📍 We nah play! We’re making the necessary repairs to restore your power. While you may not see damage to JPS structures in your area, the problem could be much further up the road. Our teams are in their hometowns working to restore power. Thank you for your patience🤝🏾

JPS

31,943 просмотров • 2 лет назад

Cooking is not difficult! 🤩 With little guide & practice, you can master your skills and impress everyone. A Big THANK YOU to everyone that has registered for the cooking class Community 🤸‍♀️ Giveaway: we have 5 free slots for the class!! If interested, type * I want* 👇

Cooking is not difficult! 🤩 With little guide & practice, you can master your skills and impress everyone. A Big THANK YOU to everyone that has registered for the cooking class Community 🤸‍♀️ Giveaway: we have 5 free slots for the class!! If interested, type * I want* 👇

Tspices Kitchen

186,663 просмотров • 3 лет назад

In this powerful video, Rhonda Byrne explains why your dreams may not be manifesting and why letting go could be the key to making them a reality. 🌟✨ Sometimes, the more we grip tightly onto our desires, the more resistance we create. By surrendering and trusting the Universe, we create the space for our dreams to manifest effortlessly. 🌌 When you let go of the “how” and “when,” you allow the Universe to work in ways that are beyond your imagination. ✨ Check out The Secret Manifestation Cards for powerful tools to manifest your best life: Ready to manifest your dreams? Leave a 🙏 in the comments! #RhondaByrne #TheSecret #lawofattraction #loa #visualization #manifestation #askbelievereceive

In this powerful video, Rhonda Byrne explains why your dreams may not be manifesting and why letting go could be the key to making them a reality. 🌟✨ Sometimes, the more we grip tightly onto our desires, the more resistance we create. By surrendering and trusting the Universe, we create the space for our dreams to manifest effortlessly. 🌌 When you let go of the “how” and “when,” you allow the Universe to work in ways that are beyond your imagination. ✨ Check out The Secret Manifestation Cards for powerful tools to manifest your best life: Ready to manifest your dreams? Leave a 🙏 in the comments! #RhondaByrne #TheSecret #lawofattraction #loa #visualization #manifestation #askbelievereceive

The Secret

18,822 просмотров • 3 месяцев назад

Here we go, I’ve been saying this for two years now: we are going to lose our jobs. („And not everyone will like it“) Sam Altman has now confirmed this fact in a Video confirmed once again. There will be a significant change in the world, at the latest when robots become established in the world of work. (Via Reddit)

Here we go, I’ve been saying this for two years now: we are going to lose our jobs. („And not everyone will like it“) Sam Altman has now confirmed this fact in a Video confirmed once again. There will be a significant change in the world, at the latest when robots become established in the world of work. (Via Reddit)

Chubby♨️

75,679 просмотров • 1 год назад

Beijing E-Town Humanoid Robot Half-Marathon is tomorrow, April 19, 2026. - Bipedal only - “We are not a car!” - Over 300 humanoid robots from 76+ teams (nearly 5× last year) - ~40% will run fully autonomous - Height limit: 2 ft 6 in – 5 ft 11 in Key Rules - 20% penalty added to finish time (×1.2), if not autonomous - Time penalties for battery swaps or full robot replacements The event runs on the same 21.1 km (13.1 miles) course as the human half-marathon, with humans and robots separated by barriers. Gimpse of practice runs:

Beijing E-Town Humanoid Robot Half-Marathon is tomorrow, April 19, 2026. - Bipedal only - “We are not a car!” - Over 300 humanoid robots from 76+ teams (nearly 5× last year) - ~40% will run fully autonomous - Height limit: 2 ft 6 in – 5 ft 11 in Key Rules - 20% penalty added to finish time (×1.2), if not autonomous - Time penalties for battery swaps or full robot replacements The event runs on the same 21.1 km (13.1 miles) course as the human half-marathon, with humans and robots separated by barriers. Gimpse of practice runs:

The Humanoid Hub

67,952 просмотров • 3 месяцев назад

Late last night a large number of our signs were vandalized across the riding. We are working to fix them now but will have to purchase more to replace many. This continued and targeted vandalism against us in particular has to stop. We will not be intimated or slow down. I will continue to persevere for you. Team #OttawaSouth

Late last night a large number of our signs were vandalized across the riding. We are working to fix them now but will have to purchase more to replace many. This continued and targeted vandalism against us in particular has to stop. We will not be intimated or slow down. I will continue to persevere for you. Team #OttawaSouth

Blair Turner

43,004 просмотров • 1 год назад

We are excited to share our #CORL2024 paper (oral) on "Learning Quadruped Locomotion Using Differentiable Simulation" done in collaboration with Sangbae Kim Massachusetts Institute of Technology (MIT). We present a new way to learn to walk in minutes without parallelization, outperforming PPO in sample efficiency! PDF: Video: We present a new framework for learning quadruped locomotion. By leveraging differentiable simulation for policy optimization, our approach achieves fast convergence and stable training, significantly outperforming model-free #ReinforcementLearning methods like PPO in sample efficiency. The key enabler is to combine a high-fidelity, non-differentiable simulator for forward dynamics with a simplified surrogate model for gradient backpropagation. Our framework enables learning quadruped walking in simulation in minutes without parallelization. When augmented with GPU parallelization, our approach allows the quadruped robot to master diverse locomotion skills on challenging terrains in minutes. This work highlights one of the first successful real-world applications of differentiable simulation for quadruped robots, offering a compelling alternative to traditional RL methods. Kudos to Yunlong Song! UZH Science University of Zurich UZH Space Hub UZH IfI European Research Council (ERC) Massachusetts Institute of Technology (MIT)MechE

We are excited to share our #CORL2024 paper (oral) on "Learning Quadruped Locomotion Using Differentiable Simulation" done in collaboration with Sangbae Kim Massachusetts Institute of Technology (MIT). We present a new way to learn to walk in minutes without parallelization, outperforming PPO in sample efficiency! PDF: Video: We present a new framework for learning quadruped locomotion. By leveraging differentiable simulation for policy optimization, our approach achieves fast convergence and stable training, significantly outperforming model-free #ReinforcementLearning methods like PPO in sample efficiency. The key enabler is to combine a high-fidelity, non-differentiable simulator for forward dynamics with a simplified surrogate model for gradient backpropagation. Our framework enables learning quadruped walking in simulation in minutes without parallelization. When augmented with GPU parallelization, our approach allows the quadruped robot to master diverse locomotion skills on challenging terrains in minutes. This work highlights one of the first successful real-world applications of differentiable simulation for quadruped robots, offering a compelling alternative to traditional RL methods. Kudos to Yunlong Song! UZH Science University of Zurich UZH Space Hub UZH IfI European Research Council (ERC) Massachusetts Institute of Technology (MIT)MechE

Davide Scaramuzza

15,533 просмотров • 1 год назад

Please use our LinkTree as a resource for donating. Browse 100+ funds for Gaza, downloadable flyers, adopt a fund, & more! Add it to your social media bios. Share it with friends & family. We’ve put in way too much work for yall to not be using this!!!

Please use our LinkTree as a resource for donating. Browse 100+ funds for Gaza, downloadable flyers, adopt a fund, & more! Add it to your social media bios. Share it with friends & family. We’ve put in way too much work for yall to not be using this!!!

Palestine Asdiqa أصدقاء فلسطين

117,013 просмотров • 1 год назад

So we did a bunch of projects with real world reinforcement learning - but it was often too inefficient to be practical to train tabula rasa. This suggests we need better priors, but acquiring these from on-robot data can often be expensive as well. In our recent work, we show that despite being fundamentally inaccurate, simulation can guide provide a cheap way to guide real-world RL finetuning to be super efficient! We propose Simulation-Guided Fine-Tuning (SGFT) - a simple paradigm for sim2real finetuning that uses simulation to provide reward shaping that accelerates real world RL finetuning *beyond* just providing an initialization. TLDR: Use value functions from sim to shape rewards for real-world RL, see large sample efficiency improvements 🧵(1/6)

So we did a bunch of projects with real world reinforcement learning - but it was often too inefficient to be practical to train tabula rasa. This suggests we need better priors, but acquiring these from on-robot data can often be expensive as well. In our recent work, we show that despite being fundamentally inaccurate, simulation can guide provide a cheap way to guide real-world RL finetuning to be super efficient! We propose Simulation-Guided Fine-Tuning (SGFT) - a simple paradigm for sim2real finetuning that uses simulation to provide reward shaping that accelerates real world RL finetuning beyond just providing an initialization. TLDR: Use value functions from sim to shape rewards for real-world RL, see large sample efficiency improvements 🧵(1/6)

Abhishek Gupta

13,637 просмотров • 1 год назад

A “Stop Trump” protest is underway in Nuuk, Greenland’s capital, led by the PM and other political leaders. "Listen to us. Our country is not for sale. We do not want to be Americans... We are Kalaallit." "Not to take over. Not to acquire. Not by pressure. Not now. Not ever."

A “Stop Trump” protest is underway in Nuuk, Greenland’s capital, led by the PM and other political leaders. "Listen to us. Our country is not for sale. We do not want to be Americans... We are Kalaallit." "Not to take over. Not to acquire. Not by pressure. Not now. Not ever."

Adam Schwarz

20,946 просмотров • 6 месяцев назад