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.

A small but powerful modification to the GELLO handle: mounting it at a 45° angle allows both the robot and the operator to be in a natural position while collecting data. .STL and .STEP in GitHub repo (🧵)

Jannik Grothusen

3,397 subscribers

42,924 views • 10 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

Turn any GitHub repo into a visual, interactive diagram in one click 🤯 Just replace `github` with `gitdiagram` in the URL… or go to: → - It maps your entire repo - Shows all files - Reveals every dependency 100% open-source and free. Repo in 🧵↓

Turn any GitHub repo into a visual, interactive diagram in one click 🤯 Just replace `github` with `gitdiagram` in the URL… or go to: → - It maps your entire repo - Shows all files - Reveals every dependency 100% open-source and free. Repo in 🧵↓

Charly Wargnier

15,034 views • 9 months ago

New project idea but left the laptop at home? 😬 Create a repo right from your phone. Name it, set visibility, and adjust the details in the GitHub Mobile app. 📱

New project idea but left the laptop at home? 😬 Create a repo right from your phone. Name it, set visibility, and adjust the details in the GitHub Mobile app. 📱

GitHub

53,028 views • 3 months ago

In the era of long-context LLMs, I needed a way to quickly convert a GitHub repo into a text file to feed as a prompt. Introducing RepoToTextForLLMs 📖 Not only does it return the full repo in one file, but it also appends a super prompt for analysis and understanding 👇

In the era of long-context LLMs, I needed a way to quickly convert a GitHub repo into a text file to feed as a prompt. Introducing RepoToTextForLLMs 📖 Not only does it return the full repo in one file, but it also appends a super prompt for analysis and understanding 👇

Pietro Schirano

104,922 views • 2 years ago

This is basically OpenAI's Operator, but free. by Browserbase. Describe tasks in plain english and it runs through the web clicking around, finding answers, and collecting data. Soon there will be no limit to these things.

This is basically OpenAI's Operator, but free. by Browserbase. Describe tasks in plain english and it runs through the web clicking around, finding answers, and collecting data. Soon there will be no limit to these things.

AI Breakfast

68,349 views • 1 year ago

Robot policies must be both reliable and highly capable to be useful; the best way to achieve this level of performance is with reinforcement learning. However, for reinforcement learning you are usually stuck between two difficult options: reinforcement in the real world is often risky and expensive, while reinforcement learning in a traditional simulator takes a lot of engineering work and has a persistent sim-to-real gap. What if instead you could train your robot purely in a world model? RISE by Jiazhi Yang et al. uses a compositional world model to predict the future and evaluate progress. This allows for a self-improving pipeline, which learns a world model from real data and then learns how the robot should perform different tasks. This pipeline results in a data-driven way to improve policy performance from real data but without real-world reinforcement learning. Watch Episode #86 of RoboPapers, with Chris Paxton and Jiafei Duan, to learn more!

Robot policies must be both reliable and highly capable to be useful; the best way to achieve this level of performance is with reinforcement learning. However, for reinforcement learning you are usually stuck between two difficult options: reinforcement in the real world is often risky and expensive, while reinforcement learning in a traditional simulator takes a lot of engineering work and has a persistent sim-to-real gap. What if instead you could train your robot purely in a world model? RISE by Jiazhi Yang et al. uses a compositional world model to predict the future and evaluate progress. This allows for a self-improving pipeline, which learns a world model from real data and then learns how the robot should perform different tasks. This pipeline results in a data-driven way to improve policy performance from real data but without real-world reinforcement learning. Watch Episode #86 of RoboPapers, with Chris Paxton and Jiafei Duan, to learn more!

RoboPapers

38,334 views • 2 months ago

STOP using the crossover step & START using a drop step! The crossover step is slower, it starts the athlete in a horrible position for acceleration, and takes away from the athletes natural athleticism.

STOP using the crossover step & START using a drop step! The crossover step is slower, it starts the athlete in a horrible position for acceleration, and takes away from the athletes natural athleticism.

Alex Simone

43,641 views • 5 months ago

Collecting robot data at scale is key to deploying working manipulation policies, and the team from Tutor Intelligence Tutor is here to tell us about how to accomplish it. Their new announcement: a massive, 100-robot “data factory,” with a behind-the-scenes look at how to build a teleoperation platform and how to make robots and policies that are useful for their customers. Tutor Intelligence is a full-stack robotics company: they build robot arms, they sell robot arms, they write the software and they train neural networks. Josh Gruenstein, Jesse Michel, Shiraz, and Joe McCalmon, and Joe McCalmon join us to tell us more about how they scale both teleop data and human interventions from their teleoperators in order to train the policies they need. Watch Episode #85 of RoboPapers, with Chris Paxton and Jiafei Duan, to learn more!

Collecting robot data at scale is key to deploying working manipulation policies, and the team from Tutor Intelligence Tutor is here to tell us about how to accomplish it. Their new announcement: a massive, 100-robot “data factory,” with a behind-the-scenes look at how to build a teleoperation platform and how to make robots and policies that are useful for their customers. Tutor Intelligence is a full-stack robotics company: they build robot arms, they sell robot arms, they write the software and they train neural networks. Josh Gruenstein, Jesse Michel, Shiraz, and Joe McCalmon, and Joe McCalmon join us to tell us more about how they scale both teleop data and human interventions from their teleoperators in order to train the policies they need. Watch Episode #85 of RoboPapers, with Chris Paxton and Jiafei Duan, to learn more!

RoboPapers

36,041 views • 2 months ago

GeoLibre has a new plugin for visualizing origin-destination data. It allows you to seamlessly explore mobility, migration, and other flow data with high-performance WebGL rendering. A big thank you to Kai Erlenbusch for developing and contributing the plugin to the GeoLibre plugin marketplace. GitHub: If you are interested in developing a GeoLibre plugin, check out the documentation: #opensource #GeoLibre

GeoLibre has a new plugin for visualizing origin-destination data. It allows you to seamlessly explore mobility, migration, and other flow data with high-performance WebGL rendering. A big thank you to Kai Erlenbusch for developing and contributing the plugin to the GeoLibre plugin marketplace. GitHub: If you are interested in developing a GeoLibre plugin, check out the documentation: #opensource #GeoLibre

Qiusheng Wu

20,871 views • 1 month ago

Steve Kerr talked about his commitment to the starting lineup, a combination that has actually been negative: “I know it hasn’t been great but it allows Draymond to start at the four… while the numbers may not be that impressive. It allows the game unfold in a way we like.”

Steve Kerr talked about his commitment to the starting lineup, a combination that has actually been negative: “I know it hasn’t been great but it allows Draymond to start at the four… while the numbers may not be that impressive. It allows the game unfold in a way we like.”

Kenzo Fukuda

88,457 views • 7 months ago

Robot-free demos can be collected through our VR interface, inspected, trained, and evaluated in a closed loop. Cool thing is that a small amount of real-robot data mixed with large-scale robot-free data can reach comparable performance, while reducing real-robot data needs by up to 20x. Code: Paper:

Robot-free demos can be collected through our VR interface, inspected, trained, and evaluated in a closed loop. Cool thing is that a small amount of real-robot data mixed with large-scale robot-free data can reach comparable performance, while reducing real-robot data needs by up to 20x. Code: Paper:

X Square Robot

175,588 views • 2 months ago

🤖 Robot takes an epic tumble into the judges’ table at the World Humanoid Robot Games The iron athlete tried to lift a weight — but couldn’t quite handle it, lost its balance and came crashing down. The mishap happened during the weightlifting final at the World Humanoid Robot Games in Beijing. After the spectacular wipeout, the poor robot was placed on a stretcher and carried away from the arena.

🤖 Robot takes an epic tumble into the judges’ table at the World Humanoid Robot Games The iron athlete tried to lift a weight — but couldn’t quite handle it, lost its balance and came crashing down. The mishap happened during the weightlifting final at the World Humanoid Robot Games in Beijing. After the spectacular wipeout, the poor robot was placed on a stretcher and carried away from the arena.

NEXTA

39,419 views • 2 days ago

Collecting dexterous humanoid robot data is difficult to scale. That's why Mengda Xu and Han Zhang built DexUMI: a tool for demonstrating how to control a dexterous robot hand, which allows you to quickly collect task data. Co-hosted by Michael Cho - Rbt/Acc and Chris Paxton

Collecting dexterous humanoid robot data is difficult to scale. That's why Mengda Xu and Han Zhang built DexUMI: a tool for demonstrating how to control a dexterous robot hand, which allows you to quickly collect task data. Co-hosted by Michael Cho - Rbt/Acc and Chris Paxton

RoboPapers

26,344 views • 1 year ago

Ever opened a repo thinking “this’ll just be a README”, and suddenly you’re building a full Python stack? Same. Get to know the hosts of our season 2 GitHub Podcast, and find out their favorite episodes from last season. Catch up on all the episodes wherever you listen to podcasts or at What’s the most unexpectedly big thing you’ve built from a small idea? 👇

Ever opened a repo thinking “this’ll just be a README”, and suddenly you’re building a full Python stack? Same. Get to know the hosts of our season 2 GitHub Podcast, and find out their favorite episodes from last season. Catch up on all the episodes wherever you listen to podcasts or at What’s the most unexpectedly big thing you’ve built from a small idea? 👇

GitHub

61,816 views • 4 days ago

ExplainThisRepo is live 🥳 A CLI package tool that explains any GitHub repo in plain English. It fetches a repo’s README + metadata and generates a human readable file in one command. try it: `pip install explainthisrepo explainthisrepo owner/repo` A thread 🧵

ExplainThisRepo is live 🥳 A CLI package tool that explains any GitHub repo in plain English. It fetches a repo’s README + metadata and generates a human readable file in one command. try it: `pip install explainthisrepo explainthisrepo owner/repo` A thread 🧵

Caleb Wodi | Calebthenics

14,396 views • 7 months ago

In my experience, robot 'generalists' are often jacks of all trades but masters of none. In training across multiple tasks and environments, robot policies fail to generalize robustly and effectively to each particular test setting. What if at test time, we non-parametrically *retrieved* “relevant” data from the training set and used it to significantly improve the performance of few-shot imitation learning to be robust to various test time scenes. Notably, we are *not* collecting lots of new data, just training more on sub-components of the same training data! Now, we’re certainly not the first to suggest retrieval, but in our new work - STRAP, we show how retrieving relevant *sub-trajectories* from offline datasets can significantly increase data reuse across tasks, when paired with an appropriate metric space. A 🧵 (1/7)

In my experience, robot 'generalists' are often jacks of all trades but masters of none. In training across multiple tasks and environments, robot policies fail to generalize robustly and effectively to each particular test setting. What if at test time, we non-parametrically retrieved “relevant” data from the training set and used it to significantly improve the performance of few-shot imitation learning to be robust to various test time scenes. Notably, we are not collecting lots of new data, just training more on sub-components of the same training data! Now, we’re certainly not the first to suggest retrieval, but in our new work - STRAP, we show how retrieving relevant sub-trajectories from offline datasets can significantly increase data reuse across tasks, when paired with an appropriate metric space. A 🧵 (1/7)

Abhishek Gupta

12,045 views • 1 year ago

The grocery robots from Starship Technologies has been getting into shenanigans in Finland now for a while. Even small towns have them and they have a range of 4kms. This one robot decided to drive off a bridge and was later seen floating in the river until a boat picked it up and returned it to the store.

The grocery robots from Starship Technologies has been getting into shenanigans in Finland now for a while. Even small towns have them and they have a range of 4kms. This one robot decided to drive off a bridge and was later seen floating in the river until a boat picked it up and returned it to the store.

Birch Brother 🪓

51,298 views • 1 month ago

In December 2006, Honda's humanoid robot ASIMO malfunctioned during a demonstration in Tokyo, Japan. While attempting to ascend a set of stairs to showcase its agility, the robot lost its balance and fell backward onto the floor. Technicians immediately rushed the stage and placed a privacy screen around the fallen robot to block the audience's view while they retrieved it.

In December 2006, Honda's humanoid robot ASIMO malfunctioned during a demonstration in Tokyo, Japan. While attempting to ascend a set of stairs to showcase its agility, the robot lost its balance and fell backward onto the floor. Technicians immediately rushed the stage and placed a privacy screen around the fallen robot to block the audience's view while they retrieved it.

Dr. Lemma

111,249 views • 8 months ago