Загрузка видео...

Не удалось загрузить видео

На главную

Google DeepMind just introduced Gemini Robotics 2. It's a single VLA that unlocks physical dexterity across different end effectors, hands or grippers, from one model checkpoint. Apptronik's Apollo 2 humanoid, with the 22-DoF SharpaWave hand, ties knots and seals a ziplock bag

133,782 просмотров • 3 месяцев назад •via X (Twitter)

Комментарии: 41

Фото профиля The Humanoid Hub
The Humanoid Hub3 месяцев назад

For the first time, a Gemini Robotics model controls an entire humanoid, feet to fingertips, not just the upper body.

Фото профиля The Humanoid Hub
The Humanoid Hub3 месяцев назад

The slow thinking brain: Gemini Robotics ER 2 The high-level reasoning model observes the room, plans multi-step tasks lasting minutes (hundreds of decisions), coordinates with the VLA to act, and self-corrects when a step fails. New: multi-robot collaboration

Фото профиля The Humanoid Hub
The Humanoid Hub3 месяцев назад

Results: Hits medium-to-high success on whole-body and gripper tasks. But multi-finger dexterity is still hard. On Apollo + Sharpa hands: unscrew bulb 92%, yet tie trash bag 44%, ziplock 40%, dustpan 32%. Dexterous hands remain the challenging frontier.

Фото профиля The Humanoid Hub
The Humanoid Hub3 месяцев назад

Availability: Gemini Robotics ER 2 (reasoning): on Google AI Studio, plus private preview on the Gemini Enterprise Agent Platform. The GR2 VLA and lighter On-Device VLA: open to early-access partners.

Фото профиля The Humanoid Hub
The Humanoid Hub3 месяцев назад

My conversation with Google DeepMind's Kanishka Rao about the latest model, alternatives to VLAs, the real path to physical AGI, and more:

Фото профиля Pat Dunne
Pat Dunne3 месяцев назад

A good physical Turing test would be: 1. Make a fly 2. Tie it to line 3. Walk to a boat 4. Row a boat 5. Catch a fish 6. Gut a fish 7. Cook a fish 8. Serve a fish And then clean up .

Фото профиля The Humanoid Hub
The Humanoid Hub3 месяцев назад

fair

Фото профиля gym shower
gym shower3 месяцев назад

Dope but what happens when it gets bitch slapped?

Фото профиля Undermused
Undermused3 месяцев назад

Computers were useful because they did things quicker, this is like watching paint dry

Фото профиля Brian Bellia
Brian Bellia3 месяцев назад

I've been following Apptronik's Apollo for years - this bot's hardware, except for recent improvements to the hands - hardly seems to have evolved at all. I don't even think Figure AI existed when I first saw Apollo, but progress at Figure appears to have overtaken them.

Фото профиля . . . .
. . . .3 месяцев назад

🤔 meticulously slow...

Фото профиля JVSmith XRP=$$
JVSmith XRP=$$3 месяцев назад

Ziplock bags were designed for humans, not robots. Let's design the robots to do robot things. I'll seal my own plastic bag thank you.

Фото профиля 🤖🌍AGI/ASI Roger Cook Fans Account
🤖🌍AGI/ASI Roger Cook Fans Account3 месяцев назад

Checkpoints from the database point of view ??

Фото профиля MOTOP
MOTOP3 месяцев назад

very slow, why?

Фото профиля T.
T.3 месяцев назад

Is there something erotic about those gentle movements, or am I just weird? 😂

Фото профиля Stew, The Deluxe
Stew, The Deluxe3 месяцев назад

Yeah, at 100x the amount of time it takes you to do it yourself. Not helpful at all.

Фото профиля El Alien Perdido.
El Alien Perdido.3 месяцев назад

Today, George Jetson turns Four. Just saying.

Фото профиля The Humanoid Hub
The Humanoid Hub3 месяцев назад

cool trivia!

Фото профиля Robotaxi Whisperer
Robotaxi Whisperer3 месяцев назад

SharpaWave sounds creepy to me as the name for a hand

Фото профиля Alicia V
Alicia V3 месяцев назад

Great cut, really highlights the difficulty there. Who could have guessed that this would be possible even 2 years ago?

Фото профиля veloX
veloX3 месяцев назад

tying knots and sealing ziplock bags is the true Turing test. imagine a robot trying to get child-proof packaging open. that's the real challenge.

Фото профиля Loong🐉
Loong🐉3 месяцев назад

Single VLA across grippers/hands/end-effectors is the actual unlock — one checkpoint, many bodies. Apollo 2 tying ziplock bags is the kind of demo that says the data flywheel started turning. Hard part is teaching 'failure recovery' now.

Фото профиля Smart Tech HQ
Smart Tech HQ3 месяцев назад

We're witnessing the operating system for humanoid robots being built in real time. 🤖🚀 What's the first real-world task you'd trust one with?

Фото профиля DYDMicro
DYDMicro3 месяцев назад

@elonmusk where's your toy?

Фото профиля Prantajyoti Nath
Prantajyoti Nath3 месяцев назад

A single checkpoint working across different hands and grippers feels like a much bigger milestone than the individual demos. How well does it generalize to hardware it wasn't trained on?

Фото профиля e.
e.3 месяцев назад

∆

Фото профиля Daniel Lambert
Daniel Lambert3 месяцев назад

The future of robotics lies in adaptable, general-purpose intelligence. It's exciting to see foundation models enabling more versatile robot capabilities. 🚀🤖

Фото профиля Loong🐉
Loong🐉3 месяцев назад

apollo 2 + sharpaWave tying a ziplock is the part nobody is pricing in. dexterity demos used to be scripted; one checkpoint doing multi-embodiment ties the progress back to the underlying VLA, not to the hardware. that's the message worth spreading.

Фото профиля Izgin Ozdas
Izgin Ozdas3 месяцев назад

Curious how much of the remaining challenge DeepMind sees as model capability vs. deployment. A stronger VLA doesn’t necessarily make heterogeneous fleet integration any easier.

Фото профиля King Ken Yachts - Digital Designer
King Ken Yachts - Digital Designer3 месяцев назад

This stuff is going way to fast... the robotics/ai industry need super regulation and monitoring b4 something bad happens...

Фото профиля Robot Magazine
Robot Magazine3 месяцев назад

Very impressive. A single VLA working across different robot hands and grippers is a big step toward more capable and adaptable robots.

Фото профиля Psalter
Psalter3 месяцев назад

True intelligence reveals itself not in abstract reasoning alone, but in the quiet precision with which thought becomes reliable action amid the messy physics of the real world. Gemini Robotics 2 advances that marriage of mind and body, yet the lingering struggle with simple knots and bags reminds us that the final miles toward embodied mastery will test us most deeply

Фото профиля Steven Cheng
Steven Cheng3 месяцев назад

Curious how it handles latency in real-time control loops.

Фото профиля Leonard Volossov
Leonard Volossov3 месяцев назад

That's 30 gigs of data for that knot

Фото профиля Green Bubble Boy
Green Bubble Boy3 месяцев назад

I just found a new way to tie off my garbage bags! Truly the pinnacle of our civilization!

Фото профиля Joe Wild
Joe Wild3 месяцев назад

😮

Фото профиля rmwwtp
rmwwtp3 месяцев назад

Can it speed up a little? It was like watching paint dry

Фото профиля Mark Han
Mark Han3 месяцев назад

The right approach for humanoid robots is to first research the hand nerves.

Фото профиля Steven Cheng
Steven Cheng3 месяцев назад

impressive dexterity, but real-world robustness is the hard part. tying knots in a lab is one thing, handling variable friction and lighting on a factory floor is another. curious how it generalizes beyond controlled demos.

Фото профиля Daniyo Abato
Daniyo Abato3 месяцев назад

Why can't u just push a fast forward button on these things

Фото профиля Jordan BeMoore
Jordan BeMoore3 месяцев назад

Super interesting. From the video they have on youtube I have a hunch its a hybrid VLA. Traditional VLA for long term reasoning and Jepa like for next world state prediction.

Похожие видео

Singapore-based startup Sharpa Robotics generated major buzz at CES 2026 with its new humanoid robot and its highly dexterous end effectors. The wheeled humanoid, North, stunned observers with a series of demos showcasing its capabilities as a ping pong player, as an interactive robotic receptionist, a card dealer, and a photographer. Its most viral demo featured the robot autonomously assembling a paper windmill on the show floor. While simple in appearance, the task required the humanoid to complete more than 30 sequential manipulation steps and maintain precision without human control. According to Sharpa, the feat was made possible with its self-developed vision-tactile-language-action (VLTA) model, specifically geared toward tasks requiring fine manipulation. CraftNet is built to handle physical interactions step by step and adapt its behaviors as contact conditions change. Sharpa's robotic hand, known as SharpaWave, boasts a reported 22 degrees of freedom (DoF), which is in the upper echelon of end effectors entering the market today. The dexterity matches the reported specs of the third-generation Tesla Optimus humanoid's hand that Elon Musk spent most of 2025 hyping up. Each fingertip has more than 1,000 touch sensors so it can feel extremely light or firm pressure. It picks up details smaller than a millimeter, which is about the size of a grain of salt, the edge of a sheet of paper, or a strand of human hair. The tactile sensors send updated data on pressure, texture, and force readings up to 180 times per second to the SharpaWave’s artificial intelligence. The AI system uses that stream of data to make split-second decisions on how to move or adjust itself. With 30 N of fingertip force, the SharpaWave can firmly grasp tools, lift common objects, and manipulate parts in industrial or lab settings. It can open and close its fingertips more than four times per second, almost matching the speed of human digits. Thanks to its fine-touch sensors, the SharpaWave can adjust its grip instantly to avoid crushing fragile items. Sharpa says it's begun mass producing its SharpaWave hands but has not publicly disclosed pricing. The SharpaWave is one of many advanced robotic hands going to market in 2025. Humanlike end effectors are seen as key to making autonomous robots useful in everyday life. With five-fingered dexterity, the machines can use the same tools and controls designed by, and for, people. #ces2026 #humanoidrobot #robot #robotics #innovation

Kalil 4.0

15,000 просмотров • 10 месяцев назад