Caydley

[DN_DEGEN @DN_degen]

THIS TINY 35-GRAM ROBOT GOT THROWN 10 METERS... AND JUST KEPT HOPPING. 🤖🦘

It's a 35-gram mini drone — about as heavy as 7 AA batteries — with just one leg. No wheels, no extra motors.

In this clip, researchers throw it 10 meters across the room to test how tough it is. It bounces, resets itself in milliseconds, and starts hopping again like nothing happened.

The secret is in that one leg. It's just a spring, not a motor. On landing, the spring compresses and fires back, launching the robot up again — like a pogo stick. That trick recovers about 90% of the landing energy, and hopping this way uses only about 25% of the energy regular flying would.

Each hop only covers a few meters, not a long trip — this isn't a cargo drone.

But that's the whole point. It's built to go where people and bigger drones can't: collapsed buildings, narrow pipes, smoky rooms during a fire.

Someday, it could even hop around on the Moon, where regular propellers don't work at all.

[Caydley @CuberPort70806]

CHINA IS LETTING HUMANOID ROBOTS TAKE CONTROL OF THE ROADS 🤖🚦

Forget crossing guards and traffic cops.

In several Chinese cities, robotic officers are already being tested on real streets — watching traffic, spotting violations and warning drivers through voice commands.

One of the Hangzhou units is nearly 1.9 meters tall and weighs around 98 kg. Its cameras and radar help it monitor the road, while its software allows it to perform eight different traffic-directing gestures in sync with the lights.

And this isn’t some distant experiment.

15 of these robots were deployed across Hangzhou in May 2026.

Apparently, the next person telling you to slow down might not be a person at all.

[Caydley @CuberPort70806]

CHINA IS LETTING HUMANOID ROBOTS TAKE CONTROL OF THE ROADS 🤖🚦

Forget crossing guards and traffic cops.

In several Chinese cities, robotic officers are already being tested on real streets — watching traffic, spotting violations and warning drivers through voice commands.

One of the Hangzhou units is nearly 1.9 meters tall and weighs around 98 kg. Its cameras and radar help it monitor the road, while its software allows it to perform eight different traffic-directing gestures in sync with the lights.

And this isn’t some distant experiment.

15 of these robots were deployed across Hangzhou in May 2026.

Apparently, the next person telling you to slow down might not be a person at all.

[Fox_AI @EchoCast_]

TESLA OPTIMUS JUST GOT THE BEST SEAT IN AN ALIEN INVASION

A humanoid robot sitting on top of a tank.

A giant unidentified object hanging in the sky.

And a scene that looks more like science fiction than reality.

That's why this footage is spreading so fast online.

What stands out isn't the tank or the explosion. It's the robot. Instead of acting like a combat unit, it looks more like an observer that was given the perfect view of the action.

That detail highlights something important about modern humanoids.

Most advanced robots today aren't being built for battlefields. They're designed for factories, warehouses, and logistics centers. Their purpose is to move objects, operate equipment, and assist human workers.

But the technology inside them is incredibly flexible.

▸ AI vision systems can recognize people and objects in real time.

▸ Modern models can react to situations they were never explicitly trained for.

▸ The same perception system can work in many different environments.

That's why the discussion around humanoid robots keeps growing.

The hardware isn't what defines the mission.

The software does.

My take:

The biggest question in robotics isn't who builds the strongest machine. It's who controls the intelligence behind it. Because once a robot can understand the world around it, changing its role may be easier than most people realize.

[RUST ORACLE @therustoracle]

FIGURE JUST THREW ITS ROBOTS INTO MOLTEN STEEL. THEY WALKED TO THE EDGE AND JUMPED. THIS WAS THE RETIREMENT PLAN. 🤖

not a movie

not CGI

a real foundry in Finland

a 75-ton furnace full of molten steel

and a humanoid robot standing on a platform above it

workers in heat gear filming from below

then it jumps

the footage you're watching is how Figure retires a robot

these F.02 units just finished 11 months on BMW's assembly line

10-hour shifts

5 days a week

30,000+ cars built

99% accuracy

they wore down exactly where the engineers predicted they would

that wear data is already inside the next model

so why melt them?

disassembling by hand pulls engineers off the F.04 build

storing them risks leaking proprietary hardware to competitors

melting solves both problems in 20 minutes

Schwarzenegger suggested it

showed up in person

did the thumbs down into the steel

"Hasta la vista, F.02"

the metal was shipped back to the US

turned into limited-edition collectibles

a robot that built 30,000 BMWs is now sitting on someone's shelf

here's the part nobody is saying out loud

the lifecycle of a humanoid robot is now shorter than a phone contract

build it

ship it

deploy it

replace it

melt it

sell what's left

11 months from factory floor to furnace

that's not a product lifecycle

that's a subscription

[NextCore @0xNextCore]

A 19 YEAR OLD MIT STUDENT BUILT AN AI BARBERSHOP WITH ANTHROPIC AND ITS ALREADY MAKING $64000 IN PURE PROFIT

But this isnt a normal barbershop. An AI robot randomly generates a completely new hairstyle for every customer, then actually cuts it meaning nobody knows what they’re walking out with until the haircut is finished.

The concept is simple: remove the human stylist from the equation and let AI decide what happens next. Every haircut becomes a gamble, every customer becomes part of the experiment, and the result is never the same twice.

Just 3 months in, the business has reportedly generated $64000 in net profit.

This could be one of the wildest examples yet of AI turning a completely ordinary service into something people actually pay to experience.

[Hoogarden @0xHoogarden]

THIS PROVES HUMANOID ROBOTS DONT NEED FUTURISTIC TOOLS TO BE USEFUL

Two identical humanoid robots were developed by a student together with a company anthropic and the experiment is surprisingly simple.

One robot uses an electric shovel. The other uses nothing more than a completely ordinary manual shovel.

And the result? Almost identical productivity.

Thats the part people are overlooking.

The value of humanoid robots may not come from giving them the most advanced equipment imaginable. Their real advantage is that they can operate with the same basic tools humans already use today.

No massive infrastructure change. No completely redesigned workplace. No specialized equipment for every single task.

Give a humanoid robot a normal tool, put it in a familiar environment, and it can perform the same physical work.

This is where humanoid robotics gets REALLY interesting.

Because if robots can work with tools that already exist, the world doesn’t need to be rebuilt for robots.

Robots can simply enter the world we already built.

And that could make scaling Physical AI much faster than most people expect.

[Caydley @CuberPort70806]

THIS ISN’T A ROBOT TRYING TO BECOME HUMAN. IT’S A MACHINE DESIGNED TO MAKE YOUR BRAIN QUESTION WHAT YOU’RE LOOKING AT. 🤖

At first, there’s almost nothing obviously robotic about the face.

The eyes move naturally. The skin shifts. Small changes in expression happen almost imperceptibly.

Then the camera reveals the rest of the machine - metal framework, cables, actuators and mechanical limbs sitting directly beneath that realistic exterior.

And that contrast is exactly what EX-Robots has been chasing.

The Chinese company, founded in Dalian in 2013, focuses heavily on human-like interaction rather than simply building stronger or faster robots.

Inside the head, miniature motors control individual facial movements. At the same time, the company is developing AI capable of detecting emotions and responding with corresponding expressions.

But this level of realism comes with an enormous price tag.

Reported costs range from roughly $207,000 to $276,000 per robot, and producing one can take around 2-4 weeks.

Originally, EX-Robots targeted museums and exhibitions.

Now the potential uses are much broader - including healthcare, education and psychological counseling.

Robotics used to be about making machines move more like us.

Now the challenge is becoming something much stranger:

Making a machine understand the face looking back at it.

[Caydley @CuberPort70806]

THIS ISN’T A ROBOT TRYING TO BECOME HUMAN. IT’S A MACHINE DESIGNED TO MAKE YOUR BRAIN QUESTION WHAT YOU’RE LOOKING AT. 🤖

At first, there’s almost nothing obviously robotic about the face.

The eyes move naturally. The skin shifts. Small changes in expression happen almost imperceptibly.

Then the camera reveals the rest of the machine - metal framework, cables, actuators and mechanical limbs sitting directly beneath that realistic exterior.

And that contrast is exactly what EX-Robots has been chasing.

The Chinese company, founded in Dalian in 2013, focuses heavily on human-like interaction rather than simply building stronger or faster robots.

Inside the head, miniature motors control individual facial movements. At the same time, the company is developing AI capable of detecting emotions and responding with corresponding expressions.

But this level of realism comes with an enormous price tag.

Reported costs range from roughly $207,000 to $276,000 per robot, and producing one can take around 2-4 weeks.

Originally, EX-Robots targeted museums and exhibitions.

Now the potential uses are much broader - including healthcare, education and psychological counseling.

Robotics used to be about making machines move more like us.

Now the challenge is becoming something much stranger:

Making a machine understand the face looking back at it.

[Caydley @CuberPort70806]

Figure gave its humanoid robots a *Terminator*-style send-off.

Instead of simply dismantling its fleet of decommissioned F.02 robots, Figure taught them to hurl themselves—on their own—straight into a 75-ton furnace filled with white-hot molten steel.

They first perfected the jump through repeated virtual simulations.
They practiced using inflatable mattresses in San Jose.
Then, the robots were sent directly to the foundry floor—a setting they had never encountered before.

Six pours.
24 hours.
Extreme temperatures and intense electromagnetic interference.

And then...

Hasta la vista, F.02.

The craziest way to "retire" robots?

[Caydley @CuberPort70806]

Figure gave its humanoid robots a *Terminator*-style send-off.

Instead of simply dismantling its fleet of decommissioned F.02 robots, Figure taught them to hurl themselves—on their own—straight into a 75-ton furnace filled with white-hot molten steel.

They first perfected the jump through repeated virtual simulations.
They practiced using inflatable mattresses in San Jose.
Then, the robots were sent directly to the foundry floor—a setting they had never encountered before.

Six pours.
24 hours.
Extreme temperatures and intense electromagnetic interference.

And then...

Hasta la vista, F.02.

The craziest way to "retire" robots?

[DN_DEGEN @DN_degen]

THIS 960G ROBOT DOG HAS 14 JOINTS, AN 8MP CAMERA, AND COSTS $1,299. 🤖🐶

Someone reaches a palm toward it. It looks. Turns. Steps back around to face the camera, ears twitching, tail-antenna swaying.

This is Hengbot's Sirius (哮天 in Chinese) — built by a Chongqing startup founded in 2022, launched publicly June 2025 after debuting at CES. The Kickstarter blew past its $5,000 goal, raising $912,768 from 1,150 backers.

The specs: 960g, 14 degrees of freedom, 4x ARM Cortex-A53 chips running up to 5 TOPS of onboard AI, an 8MP camera, and a 1.8" LCD screen for a face that animates roughly 10 different "moods."

It walks, runs, jumps, dances, gives a paw, reacts to a head touch — but the honest limits are real: 45-50 minutes of runtime per charge, indoor flat floors only, no self-charging dock, and reviewers report it still falls on uneven ground.

No attempt at photorealism. Just a screen-faced companion robot, priced like a nice laptop.

[Clavion @clavion95]

1.25-METER WALL. 96% OF ITS OWN HEIGHT. THIS HUMANOID CLIMBS IT — THEN ROLLS OFF THE OTHER SIDE.
Run. Vault. Climb. Recover.
One move is a stunt.
Chaining all of them through the same course is a capability.
The gap between robot movement and human movement is getting very small.

[punpun @punpunism]

THIS IS WHAT COMBAT LOOKS LIKE WHEN 1 SIDE RISKS ABSOLUTELY NOTHING

Look at the targeting overlay, not the explosions. The machine holds a lock on a man lying in the mud through fog and smoke, and keeps walking while it holds it.

Clearing a trench has always been the worst job in any war. Somebody has to go around the corner first and find out what is there. The first man through takes the hit. Armies have accepted that trade for 110 years because there was no other way to do it.

A quadruped does not need to know what is around the corner. It walks in, carries a sensor that reads heat instead of light, and sends the picture back to an operator sitting 2 kilometres behind the line with a screen and a controller.

The hard part was never the weapon. It was the ground. Mud, craters, broken concrete, water up to the knee. That terrain is the reason wheels and tracks never went into trenches and the reason it stayed a human job. Legs that stay upright in that mess are a 2020s machine, not a 1990s one.

An infantry squad is 9 people. Each one has to be trained, paid, fed and buried. A machine that walks the same trench gets replaced by a purchase order.

The men in that trench are not losing to a robot. They are losing to a side that has nothing left to lose.

[Jess @404jesss]

A ROBOT TOOK A RUN-UP AND JUMPED INTO MOLTEN STEEL, AND NOBODY TRIED TO STOP IT

Watch the run-up, not the splash. Nothing alive accelerates toward 1,600 degrees.

That reflex is the whole reason the job looks the way it does. Every safety rule in a steel mill is built on the assumption that a body will stop at the edge of the heat before the brain even decides anything. Liquid steel gives no second attempt, and an aluminized suit is rated for a few seconds of radiant heat, not for contact.

So for 150 years the work at the furnace was done through distance. Long poles, long lances, short shifts, and a hard line that nobody crosses. The men in the green suits in that frame are the closest any human gets, and they are still 10 metres back behind a rail.

A machine has no edge to stop at. It does not slow down, does not look, does not weigh anything up. It runs in, does what it was sent to do, and ends as scrap at the bottom of the ladle.

That is the real unlock in heavy industry, and it has nothing to do with strength. The machine is allowed to be destroyed, and it has no instinct arguing against it.

A man stops at the heat. The machine takes a run-up.

[Caydley @CuberPort70806]

THIS ROBOT DOESN’T JUST KNOW KARATE. IT KNOWS HOW TO RESPOND. 🤖🥋

What starts as an ordinary robotics demo quickly turns into something much harder to process.

Two people move toward the robot from different directions - and instead of standing there waiting for a programmed sequence, it adjusts to what they’re doing.

It blocks.
It changes position.
It strikes back.
It keeps track of both opponents.

And that’s the part that feels strange.

The movements are no longer the slow, predictable motions we usually associate with robots. The footwork, timing and reactions make it look surprisingly close to a human sparring partner.

A few years ago, getting a humanoid robot to walk naturally was already impressive.

Now we’re seeing machines interact physically with people in real time.

The technology is moving from simply following instructions to understanding and responding to movement around it.

And that raises a much bigger question:

How far should we actually take this?

[Caydley @CuberPort70806]

THIS ROBOT DOESN’T JUST KNOW KARATE. IT KNOWS HOW TO RESPOND. 🤖🥋

What starts as an ordinary robotics demo quickly turns into something much harder to process.

Two people move toward the robot from different directions - and instead of standing there waiting for a programmed sequence, it adjusts to what they’re doing.

It blocks.
It changes position.
It strikes back.
It keeps track of both opponents.

And that’s the part that feels strange.

The movements are no longer the slow, predictable motions we usually associate with robots. The footwork, timing and reactions make it look surprisingly close to a human sparring partner.

A few years ago, getting a humanoid robot to walk naturally was already impressive.

Now we’re seeing machines interact physically with people in real time.

The technology is moving from simply following instructions to understanding and responding to movement around it.

And that raises a much bigger question:

How far should we actually take this?

[Fox_AI @EchoCast_]

This robot doesn't walk to the target. It flies there. Jetpack hover. Mid-air swing. Rifle fired in flight. Same machine. Three tests. Feet never touch the floor. Warehouse, rescue, or something darker? Where do you send it first?

[Future Memo @TheFutureMemo]

$25,600 AGIBOT HUMANOID. 30 DEGREES OF FREEDOM. AND IT STILL KNOCKS THE TABLE OVER!

The robot in this restaurant clip appears to be an AgiBot X2. During a performance, it starts moving wildly, knocks objects off the table and keeps going while staff try to stop it. The official AGIBOT store lists the X2 Neo at $25,600. That is the published price of the Neo version, not a confirmed price for the exact unit in this video.

The X2 Neo has 30 degrees of freedom, weighs about 35 kg and can operate for around 2 hours at a walking speed of 0.5 m/s. That is a lot of hardware and control packed into a robot designed for performances, interaction and human-like movement.

But this is the real challenge: making a robot perform a dance is one thing. Making it control every movement around tables, dishes and people is another. One uncontrolled swing can turn a $25,600 demonstration into a very expensive cleanup.

The next robotics breakthrough isn’t making machines move. It’s making sure they know when to stop.

[Woka @0xWoka]

Two robots on the floor😮 One still standing Guess which one did it

11:39:47, office CCTV It crouches, it lunges, and four seconds later it's strolling across the room like nothing happened