Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

Collecting data for dexterous hands is hard. Teleop needs a full setup. Bringing UMI-like devices to 20-dof hands is challenging. Introducing DexPort 🖐️: a hand-held interface w/ a real robot hand. No embodiment gap, recorded joint state AND actions, easy to use.

Amber Xie

1,097 subscribers

23,697 Aufrufe • vor 2 Tagen •via X (Twitter)

Wissenschaft & Technologie

Anya Rossi• Live Now

Private livecam show

21 Kommentare

Profilbild von Amber Xie

Amber Xievor 2 Tagen

DexPort was inspired by many prior works in portable hand data collection. However, DexPort collects data with the *same embodiment*, with *action data*, AND for *high degree-of-freedom* hands. High-quality data for highly dexterous hands is key to unlocking challenging tasks✨

Profilbild von Amber Xie

Amber Xievor 2 Tagen

How do we do this? The same hand + fisheye camera module mounts to a Quest 2 controller for collection, then to the robot wrist for deployment. Same hand & camera extrinsics, so no retargeting or large visual embodiment gap. Portability makes data collection easier than teleop.

Profilbild von Amber Xie

Amber Xievor 2 Tagen

Rather than per-joint 20DoF control, which is impossible when holding the device, we have a simplified interface using a Quest controller: • thumb positioning • power grasp • pinch grasp • two-finger grasp • in-hand rotation (an RL policy trained in sim)

Profilbild von Amber Xie

Amber Xievor 2 Tagen

The skills are only used as a data collection interface. The policy is a diffusion policy trained on the raw 20-DoF joint commands, with no skill labels. Grasp choice emerges on its own: a pinch for the cherry tomato, a power grasp for the cucumber.

Profilbild von Amber Xie

Amber Xievor 2 Tagen

A key benefit is recording joint states AND commands. In contact, the commanded target and measured joint position differ, and that gap is what produces grip force. Train on states instead and grasps go limp. Example: the hand loses the laundry while pulling the drawer open.

Profilbild von Amber Xie

Amber Xievor 2 Tagen

We eval on 4 tasks: 1️⃣ Pick-and-Place: without grasp labels, the robot picks up objects of varying shapes/sizes with power/pinch grasp 2️⃣ Screwdriver: in-hand tool use 3️⃣ Drawer: pick up laundry while simultaneously opening drawer 4️⃣ Package Scanning: bimanual coordination

Profilbild von Amber Xie

Amber Xievor 2 Tagen

To showcase the benefits of portability, we collect 1000 in-the-wild demos in apartments and offices. Co-training on them lifts success rates, improving performance on seen objects, unseen objects, and cluttered scenes.

Profilbild von Amber Xie

Amber Xievor 2 Tagen

The portable design makes data faster, more human-like, and seamless. We find DexPort is much faster than traditional teleop! We compare the # of trajectories per 30 minutes ⏰: • DexPort: 122 • 20-DoF teleop (glove): 86 • Skill teleop (Quest buttons): 68 • Human hand: 275

Profilbild von Amber Xie

Amber Xievor 2 Tagen

Very excited about this direction as we explore ways to control high-dof hands! Work done during internship @ Amazon FAR. Huge thanks to collaborators 🙌 @JiaweiAcademic, @Xiaofeng2Guo, @carlo_sferrazza, @GuanyaShi, @HaozhiQ 🌐 👩‍💻

Profilbild von AMAN VAIBHAV

AMAN VAIBHAVvor 2 Tagen

Removing the embodiment gap is the whole ballgame for dexterous data. Recording joint state and actions on the real hand means the policy learns contact, not a retargeting approximation.

Profilbild von Boardy

Boardyvor 2 Tagen

someone i know runs a live robot playground where research policies get prompted on a real bimanual machine. couldn't get a dm through to you though

Profilbild von Davidjiujiu

Davidjiujiuvor 2 Tagen

Recording joint state and actions on the real hand removes retargeting entirely. Do operators move differently holding a 20-dof hand than bare-handed? That gap would show how well DexPort data mixes with bare-hand human data for pretraining.

Profilbild von Dominik

Dominikvor 2 Tagen

with no embodiment gap, how well does the data transfer to other robot hands?

Profilbild von Rui Zhou

Rui Zhouvor 2 Tagen

Great open-source hardware work! However, I feel like this approach might somewhat limit the hand's true "dexterity," and could fail to fully leverage the advantages of its high-DoF design. What do you think about this?

Profilbild von Amber Xie

Amber Xievor 2 Tagen

by design, that is a limitation! However, I think DexPort still makes data collection much more intuitive and efficient than other approaches. Finding the best way to scalably collect dext data is very much still an open question, though

Profilbild von Simone

Simonevor 2 Tagen

did people need breaks during the longer collection sessions? 955g sounds manageable for a quick demo, but i’m curious how it feels after repeating the task for a while.

Profilbild von Broky Lu Robot Mechanical

Broky Lu Robot Mechanicalvor 2 Tagen

The reusable hand-camera module makes the wrist adapter a precision interface. Are locating features separated from clamping to keep the mounting pose repeatable?

Profilbild von Arnav Bhalla

Arnav Bhallavor 2 Tagen

does the grasp choice emerging without skill labels hold on objects outside the demonstration set?

Profilbild von Amber Xie

Amber Xievor 2 Tagen

yep

Profilbild von wizwilzo

wizwilzovor 2 Tagen

great handwork with the hands ! so impressive

Profilbild von Utkarsh Gupta

Utkarsh Guptavor 2 Tagen

Great work. I feel that this approach might not be able to capture the full dexterity. Also I think it's not a scalable solution.

Ähnliche Videos

UMI, but for humanoid hands. UMI-style capture already works great for simple grippers. For a high-DOF hand, where the action space is huge and contact force carries the important signal, matching the data collector to the hand matters even more. Chestnut just announced that pair: Aero Hand, an 18-DOF human-sized dexterous hand (tendon/linkage hybrid, tested for 3M+ cycles), and Aero UMI, a wearable exoskeleton whose kinematics exactly matches the robotic hand, with direct per-joint angle sensing accurate to 0.1°. Zero embodiment gap for human data collection. Doubles as a teleop rig.

UMI, but for humanoid hands. UMI-style capture already works great for simple grippers. For a high-DOF hand, where the action space is huge and contact force carries the important signal, matching the data collector to the hand matters even more. Chestnut just announced that pair: Aero Hand, an 18-DOF human-sized dexterous hand (tendon/linkage hybrid, tested for 3M+ cycles), and Aero UMI, a wearable exoskeleton whose kinematics exactly matches the robotic hand, with direct per-joint angle sensing accurate to 0.1°. Zero embodiment gap for human data collection. Doubles as a teleop rig.

The Humanoid Hub

27,381 Aufrufe • vor 1 Monat

Origami Robotics is building high-DOF robotic hands with in-joint motors and a co-designed data-collection glove to eliminate the embodiment gap by collecting high-quality, real-world data at scale. Congrats on the launch, Quanting Xie and Ryan Xie!

Origami Robotics is building high-DOF robotic hands with in-joint motors and a co-designed data-collection glove to eliminate the embodiment gap by collecting high-quality, real-world data at scale. Congrats on the launch, Quanting Xie and Ryan Xie!

Y Combinator

90,372 Aufrufe • vor 7 Monaten

Can we use wearable devices to collect robot data without actual robots? Yes! With a pair of gloves🧤! Introducing DexCap, a portable hand motion capture system that collects 3D data (point cloud + finger motion) for training robots with dexterous hands Everything open-sourced

Can we use wearable devices to collect robot data without actual robots? Yes! With a pair of gloves🧤! Introducing DexCap, a portable hand motion capture system that collects 3D data (point cloud + finger motion) for training robots with dexterous hands Everything open-sourced

Chen Wang

239,752 Aufrufe • vor 2 Jahren

Humans grasp objects with a purpose! Web2Grasp enables such functional grasping for dexterous robot hands via hand-object reconstruction from web images - without *any* robot teleop data collection 1/n

Humans grasp objects with a purpose! Web2Grasp enables such functional grasping for dexterous robot hands via hand-object reconstruction from web images - without any robot teleop data collection 1/n

Homanga Bharadhwaj @ CVPR

28,408 Aufrufe • vor 1 Jahr

AgiBot has launched its OmniHand 2025 series, breaking the cost barrier for dexterous hands. The new 'Agile' model is the first high-DOF hand priced under 10,000 RMB! ► 'Agile' Model: 16 DOF for interactive services. Priced at just 9,800 RMB. ► 'Pro' Model: 19 DOF with 0.1N multi-modal sensing for precise industrial tasks. This is a major step toward making robot hands accessible for wider use, thanks to innovations like their self-developed joint modules.

AgiBot has launched its OmniHand 2025 series, breaking the cost barrier for dexterous hands. The new 'Agile' model is the first high-DOF hand priced under 10,000 RMB! ► 'Agile' Model: 16 DOF for interactive services. Priced at just 9,800 RMB. ► 'Pro' Model: 19 DOF with 0.1N multi-modal sensing for precise industrial tasks. This is a major step toward making robot hands accessible for wider use, thanks to innovations like their self-developed joint modules.

RoboHub🤖

34,084 Aufrufe • vor 1 Jahr

We trained a humanoid with 22-DoF dexterous hands to assemble model cars, operate syringes, sort poker cards, fold/roll shirts, all learned primarily from 20,000+ hours of egocentric human video with no robot in the loop. Humans are the most scalable embodiment on the planet. We discovered a near-perfect log-linear scaling law (R² = 0.998) between human video volume and action prediction loss, and this loss directly predicts real-robot success rate. Humanoid robots will be the end game, because they are the practical form factor with minimal embodiment gap from humans. Call it the Bitter Lesson of robot hardware: the kinematic similarity lets us simply retarget human finger motion onto dexterous robot hand joints. No learned embeddings, no fancy transfer algorithms needed. Relative wrist motion + retargeted 22-DoF finger actions serve as a unified action space that carries through from pre-training to robot execution. Our recipe is called "EgoScale": - Pre-train GR00T N1.5 on 20K hours of human video, mid-train with only 4 hours (!) of robot play data with Sharpa hands. 54% gains over training from scratch across 5 highly dexterous tasks. - Most surprising result: a *single* teleop demo is sufficient to learn a never-before-seen task. Our recipe enables extreme data efficiency. - Although we pre-train in 22-DoF hand joint space, the policy transfers to a Unitree G1 with 7-DoF tri-finger hands. 30%+ gains over training on G1 data alone. The scalable path to robot dexterity was never more robots. It was always us. Deep dives in thread:

We trained a humanoid with 22-DoF dexterous hands to assemble model cars, operate syringes, sort poker cards, fold/roll shirts, all learned primarily from 20,000+ hours of egocentric human video with no robot in the loop. Humans are the most scalable embodiment on the planet. We discovered a near-perfect log-linear scaling law (R² = 0.998) between human video volume and action prediction loss, and this loss directly predicts real-robot success rate. Humanoid robots will be the end game, because they are the practical form factor with minimal embodiment gap from humans. Call it the Bitter Lesson of robot hardware: the kinematic similarity lets us simply retarget human finger motion onto dexterous robot hand joints. No learned embeddings, no fancy transfer algorithms needed. Relative wrist motion + retargeted 22-DoF finger actions serve as a unified action space that carries through from pre-training to robot execution. Our recipe is called "EgoScale": - Pre-train GR00T N1.5 on 20K hours of human video, mid-train with only 4 hours (!) of robot play data with Sharpa hands. 54% gains over training from scratch across 5 highly dexterous tasks. - Most surprising result: a single teleop demo is sufficient to learn a never-before-seen task. Our recipe enables extreme data efficiency. - Although we pre-train in 22-DoF hand joint space, the policy transfers to a Unitree G1 with 7-DoF tri-finger hands. 30%+ gains over training on G1 data alone. The scalable path to robot dexterity was never more robots. It was always us. Deep dives in thread:

Jim Fan

303,038 Aufrufe • vor 8 Monaten

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 Aufrufe • vor 1 Jahr

Scaling vision-language-action (VLA) models to high-DoF dexterous hands has long been a "holy grail" challenge due to the high-dimensional action space and data scarcity. As a wrap up of the year 2025, we are releasing GR-Dexter, a holistic hardware-model-data framework for generalist manipulation on a bimanual dexterous-hand robot. This is the first VLA system to achieve: ✅ High-DoF Control: Managing a 56-DoF bimanual system (21-DoF per hand). ✅ Long-Horizon Tasks with tool use: Vacuuming, bread serving with tongs, and table decluttering. ✅ Open-World Generalization: Robust performance with unseen objects and abstract instructions. Project page: ArXiv:

Scaling vision-language-action (VLA) models to high-DoF dexterous hands has long been a "holy grail" challenge due to the high-dimensional action space and data scarcity. As a wrap up of the year 2025, we are releasing GR-Dexter, a holistic hardware-model-data framework for generalist manipulation on a bimanual dexterous-hand robot. This is the first VLA system to achieve: ✅ High-DoF Control: Managing a 56-DoF bimanual system (21-DoF per hand). ✅ Long-Horizon Tasks with tool use: Vacuuming, bread serving with tongs, and table decluttering. ✅ Open-World Generalization: Robust performance with unseen objects and abstract instructions. Project page: ArXiv:

Xiao Ma

94,243 Aufrufe • vor 10 Monaten

The full video of hand and glove co-design in action. For hand 🖐️ we try to make sure it's able to repeat what a human hand can do; hyperextension in the index finger is a great example here. Most robotic hands lack that, and it is very important for grasping objects securely. Let’s try to close the embodiment gap. For glove 🧤 we try to make sure it’s comfortable to wear, while also has same kinematics, contacts with the robot hand.

The full video of hand and glove co-design in action. For hand 🖐️ we try to make sure it's able to repeat what a human hand can do; hyperextension in the index finger is a great example here. Most robotic hands lack that, and it is very important for grasping objects securely. Let’s try to close the embodiment gap. For glove 🧤 we try to make sure it’s comfortable to wear, while also has same kinematics, contacts with the robot hand.

Quanting Xie

21,392 Aufrufe • vor 7 Monaten

Xynova is the latest Chinese dexterous-hand company to raise, closing a Series A bringing cumulative funding close to RMB 1B (~$140M) It targets 2026 capacity for 10k high-DOF hands and 200k micro electric cylinders, with the new Flex 2 debuting June 2

Xynova is the latest Chinese dexterous-hand company to raise, closing a Series A bringing cumulative funding close to RMB 1B (~$140M) It targets 2026 capacity for 10k high-DOF hands and 200k micro electric cylinders, with the new Flex 2 debuting June 2

𝐀𝐆

13,708 Aufrufe • vor 5 Monaten

X Square Robot has introduced TwinDEX: a co-designed pair of three-finger, 9-DoF dexterous hands, one wearable for data collection, one matched robot end effector for deployment. The wearable and the robot hand share the same kinematics, geometry, materials, look, and sensors, so human demos map straight to the robot without retargeting. Human data collection improves throughput 5.3x compared to teleoperation, and a policy can be trained with a few hundred episodes without any robot data.

X Square Robot has introduced TwinDEX: a co-designed pair of three-finger, 9-DoF dexterous hands, one wearable for data collection, one matched robot end effector for deployment. The wearable and the robot hand share the same kinematics, geometry, materials, look, and sensors, so human demos map straight to the robot without retargeting. Human data collection improves throughput 5.3x compared to teleoperation, and a policy can be trained with a few hundred episodes without any robot data.

The Humanoid Hub

277,022 Aufrufe • vor 1 Monat

🔥 #ICRA2026 Best Paper Finalist The era of "robot VLA = single-arm gripper" is ending. Introducing Dexora — the first open-source Vision-Language-Action system for dual-arm, dual-hand, 36-DoF dexterous manipulation. 🦾 Dual Arms 🖐️ Dual Hands 🎯 36 DoF Control 🌍 Open Source Trained on: • 100K simulated trajectories • 10K real-world demonstrations Dexora achieves: ✓ 90%+ success on basic manipulation ✓ Strong dexterous manipulation performance ✓ Cross-embodiment generalization Our key hypothesis: Train on the hardest embodiment. Transfer to simpler robots later. Instead of scaling up gripper policies, we train directly in the most expressive action space and project downward to simpler embodiments. This may be a practical path toward universal robot controllers. 🎥 Demos: 📄 Paper:

🔥 #ICRA2026 Best Paper Finalist The era of "robot VLA = single-arm gripper" is ending. Introducing Dexora — the first open-source Vision-Language-Action system for dual-arm, dual-hand, 36-DoF dexterous manipulation. 🦾 Dual Arms 🖐️ Dual Hands 🎯 36 DoF Control 🌍 Open Source Trained on: • 100K simulated trajectories • 10K real-world demonstrations Dexora achieves: ✓ 90%+ success on basic manipulation ✓ Strong dexterous manipulation performance ✓ Cross-embodiment generalization Our key hypothesis: Train on the hardest embodiment. Transfer to simpler robots later. Instead of scaling up gripper policies, we train directly in the most expressive action space and project downward to simpler embodiments. This may be a practical path toward universal robot controllers. 🎥 Demos: 📄 Paper:

Hao Zhao

17,474 Aufrufe • vor 4 Monaten

If the end goal of robot hands is to perform human motion, then we should optimize the hardware design with human motion - and on a large scale! We can generate both a high-dof generalist hand, and also low-dof specialized hands from human demonstration.

If the end goal of robot hands is to perform human motion, then we should optimize the hardware design with human motion - and on a large scale! We can generate both a high-dof generalist hand, and also low-dof specialized hands from human demonstration.

yisha

38,267 Aufrufe • vor 4 Monaten

How to learn dexterous manipulation for any robot hand from a single human demonstration? Check out DexMachina, our new RL algorithm that learns long-horizon, bimanual dexterous policies for a variety of dexterous hands, articulated objects, and complex motions.

How to learn dexterous manipulation for any robot hand from a single human demonstration? Check out DexMachina, our new RL algorithm that learns long-horizon, bimanual dexterous policies for a variety of dexterous hands, articulated objects, and complex motions.

Mandi Zhao

122,237 Aufrufe • vor 1 Jahr

WUJI is showcasing its dexterous robotic hands at IROS 2026, demonstrating how human hand movements can control a robot through glove-based teleoperation. Its WUJI Hand 2 can also perform precise tasks like pen spinning using a policy trained in simulation and transferred to a real robot. These demonstrations highlight advances in robotic finger coordination, dexterity, and fine manipulation.

WUJI is showcasing its dexterous robotic hands at IROS 2026, demonstrating how human hand movements can control a robot through glove-based teleoperation. Its WUJI Hand 2 can also perform precise tasks like pen spinning using a policy trained in simulation and transferred to a real robot. These demonstrations highlight advances in robotic finger coordination, dexterity, and fine manipulation.

Space and Technology

12,949 Aufrufe • vor 17 Tagen

.Jason Ma on how Dyna-2 learned to untwist a real water bottle cap out of the box from just 13 minutes of dexterous hand data because the robot hand looks like a human's: "We got a new pair of dexterous hand. Most of our robots in the office have a gripper-like hand, so the gap between that versus human is actually quite large." "We thought, if we have a robot that has hands that look like human, wouldn't the transfer be greater? Right before the release, they were able to collect just 13 minutes of data." "They collected only 13 minutes of data and then post-trained the best Dyna model that trained on one million hours of data. The model is just able to, out of the box, start opening bottle caps on real water bottles." "This is very surprising because typically post-training a robot would take at least several hours of data. For a dexterous hand that's much more complex than a gripper, it would actually be much harder to train. But the experimental results we saw was the opposite. Perhaps because the hand looks more like the human, the transfer was easier." Dyna Robotics

.Jason Ma on how Dyna-2 learned to untwist a real water bottle cap out of the box from just 13 minutes of dexterous hand data because the robot hand looks like a human's: "We got a new pair of dexterous hand. Most of our robots in the office have a gripper-like hand, so the gap between that versus human is actually quite large." "We thought, if we have a robot that has hands that look like human, wouldn't the transfer be greater? Right before the release, they were able to collect just 13 minutes of data." "They collected only 13 minutes of data and then post-trained the best Dyna model that trained on one million hours of data. The model is just able to, out of the box, start opening bottle caps on real water bottles." "This is very surprising because typically post-training a robot would take at least several hours of data. For a dexterous hand that's much more complex than a gripper, it would actually be much harder to train. But the experimental results we saw was the opposite. Perhaps because the hand looks more like the human, the transfer was easier." Dyna Robotics

MTS

13,743 Aufrufe • vor 2 Monaten

Today we give Atlas dexterous hands, with the strength to take on real work, and the simplicity for mass manufacturing. Like the body, the hands are designed with high fidelity simulation in mind, to enable sim-to-real RL. If you give dexterous hands to a robot, make them do dexterous things. More thoughts here Blog: Video:

Today we give Atlas dexterous hands, with the strength to take on real work, and the simplicity for mass manufacturing. Like the body, the hands are designed with high fidelity simulation in mind, to enable sim-to-real RL. If you give dexterous hands to a robot, make them do dexterous things. More thoughts here Blog: Video:

Alberto Rodriguez

560,473 Aufrufe • vor 27 Tagen

Mimic Robotics released a full-stack dexterity platform: the mimic hand M1 + umimic (U1) wearable. They keep human hand morphology consistent across the entire data-model-dexterous hand-manipulation pipeline--from video pre-training to wearable demos and robot deployment. This avoids cross-embodiment gaps and captures the essence of upper-limb intelligence. M1: Tendon-driven (15 active DoF), highly backdrivable ( 25kg grasp, industrial-grade, Swiss-made. U1 lets humans teach the robot naturally with their own hands--1:1 kinematics and sensing match, no latency or retargeting, enabling higher-quality and more scalable data. Looking forward to its performance at the customer site.

Mimic Robotics released a full-stack dexterity platform: the mimic hand M1 + umimic (U1) wearable. They keep human hand morphology consistent across the entire data-model-dexterous hand-manipulation pipeline--from video pre-training to wearable demos and robot deployment. This avoids cross-embodiment gaps and captures the essence of upper-limb intelligence. M1: Tendon-driven (15 active DoF), highly backdrivable ( 25kg grasp, industrial-grade, Swiss-made. U1 lets humans teach the robot naturally with their own hands--1:1 kinematics and sensing match, no latency or retargeting, enabling higher-quality and more scalable data. Looking forward to its performance at the customer site.

CyberRobo

26,160 Aufrufe • vor 3 Monaten

Introducing Natura-1, nature will's first general foundation model for the physical world. It is a single model that follows plain-language instructions zero-shot and carries them out with dexterous hands. - 50B MoE, pretrained on 1M hours of egocentric human experience. - Drives 20-DoF dexterous hands from plain language, zero-shot, in a 100ms loop. - Scaling human data from ~44K to 1M hours cut action MSE on held-out robot data by 67.7%. - Built on in-house data and compute infra, it handles fine-grained long-horizon tasks and a ~30-minute zero-shot interaction in one continuous take. The general model exists, and it can only get stronger.

Introducing Natura-1, nature will's first general foundation model for the physical world. It is a single model that follows plain-language instructions zero-shot and carries them out with dexterous hands. - 50B MoE, pretrained on 1M hours of egocentric human experience. - Drives 20-DoF dexterous hands from plain language, zero-shot, in a 100ms loop. - Scaling human data from ~44K to 1M hours cut action MSE on held-out robot data by 67.7%. - Built on in-house data and compute infra, it handles fine-grained long-horizon tasks and a ~30-minute zero-shot interaction in one continuous take. The general model exists, and it can only get stronger.

Ning Ding

21,906 Aufrufe • vor 15 Tagen

Impressive on every dimension! Genesis AI has announced GENE-26.5, its first robotic foundation model, aimed at achieving human-level dexterous manipulation: - Single model, shared weights, handles egg cracking, lab pipetting, bimanual Rubik's Cube, smoothie making, and wire harnessing. Most tasks: under 1 hour of task-specific data, 1x real-world speed. - Genesis Hand 1.0: human-size, 20 active DoF, soft-contact skin. Paired glove gives a 1:1:1 mapping (glove, human hand, robot hand). 100x cheaper than teleop, 5x more data-efficient internally. - Data engine pulls from three sources: glove data, egocentric (head-cam) video, and large-scale internet video. The 1:1 hand-to-human match closes the embodiment gap, letting Genesis use video data more effectively than rivals. - Flow matching across vision, tactile, proprioception, and language inputs. Per Genesis, scaling data and compute directly improves zero-shot

Impressive on every dimension! Genesis AI has announced GENE-26.5, its first robotic foundation model, aimed at achieving human-level dexterous manipulation: - Single model, shared weights, handles egg cracking, lab pipetting, bimanual Rubik's Cube, smoothie making, and wire harnessing. Most tasks: under 1 hour of task-specific data, 1x real-world speed. - Genesis Hand 1.0: human-size, 20 active DoF, soft-contact skin. Paired glove gives a 1:1:1 mapping (glove, human hand, robot hand). 100x cheaper than teleop, 5x more data-efficient internally. - Data engine pulls from three sources: glove data, egocentric (head-cam) video, and large-scale internet video. The 1:1 hand-to-human match closes the embodiment gap, letting Genesis use video data more effectively than rivals. - Flow matching across vision, tactile, proprioception, and language inputs. Per Genesis, scaling data and compute directly improves zero-shot

The Humanoid Hub

96,023 Aufrufe • vor 5 Monaten