正在加载视频...

视频加载失败

返回首页

A ROBOT WALKS A WIRE OVER A 1200 M DROP, AND WHEN AN EAGLE FLIES PAST, NOTHING BREAKS ITS FOCUS The cable is a couple centimeters wide. Below the feet, nothing but air down to the valley floor. ▸ every step on the wire is constant ankle correction, since...

Shred_0x's profile picture

Shredder

4,492 subscribers

16,082 次观看 • 11 天前 •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

OPTIMUS JUMPED FROM A HELICOPTER AT 4000 METERS WITH NO INSTRUCTOR NEXT TO IT Tesla showed the humanoid in free fall Not a marketing stunt but a test of full autonomous coordination mid air ▸ exiting the helicopter with no harness backup and no second of hesitation ▸ controlled acrobatics in airflow where one degree of tilt changes the whole trajectory ▸ chute opens and the body hangs stable under the canopy with zero jerk A skydiver trains for years so the body reads the airflow on instinct The robot gets that same skill through a model that can be copied onto a thousand other frames This is not about walking across a lab floor anymore This is about a system that does not panic with four kilometers of air under its feet
0:14

OPTIMUS JUMPED FROM A HELICOPTER AT 4000 METERS WITH NO INSTRUCTOR NEXT TO IT Tesla showed the humanoid in free fall Not a marketing stunt but a test of full autonomous coordination mid air ▸ exiting the helicopter with no harness backup and no second of hesitation ▸ controlled acrobatics in airflow where one degree of tilt changes the whole trajectory ▸ chute opens and the body hangs stable under the canopy with zero jerk A skydiver trains for years so the body reads the airflow on instinct The robot gets that same skill through a model that can be copied onto a thousand other frames This is not about walking across a lab floor anymore This is about a system that does not panic with four kilometers of air under its feet

Shredder

196,230 次观看 • 19 天前

Super excited about Hydra-0 from Hongyu Li and team! The key idea is to use flow as a shared visual interface across embodiments/objects for controllable video generation, allowing a single generalist world model to learn from human, handheld-gripper, and robot interaction data. My favorite result is the video below: start from a real video of a human doing the task (left), extract the desired object flow, and condition the model on that flow (right). The model then hallucinates a plausible robot motion that could produce the same object motion. Very cool glimpse of how a generalist world model can bridge human demonstrations and robot control.
0:13

Super excited about Hydra-0 from Hongyu Li and team! The key idea is to use flow as a shared visual interface across embodiments/objects for controllable video generation, allowing a single generalist world model to learn from human, handheld-gripper, and robot interaction data. My favorite result is the video below: start from a real video of a human doing the task (left), extract the desired object flow, and condition the model on that flow (right). The model then hallucinates a plausible robot motion that could produce the same object motion. Very cool glimpse of how a generalist world model can bridge human demonstrations and robot control.

Yunzhu Li

10,850 次观看 • 1 个月前

THIS ROBOT HAS A DEAD FLY'S BRAIN AND IT WALKS LIKE A FLY scientists sliced the fly into thousands of layers thinner than a virus to copy every wire in its body the fly didn't survive the map did: 166,700 neurons, from the brain all the way down to the legs so somebody built it a new body six metal legs, 18 motors, two camera eyes and a battery strapped to its back no walking code, they just wired the leg neurons into the motors and switched it on > it turned its head and looked straight at the camera > then it walked, three legs down, three legs up, the exact way a real fly does > when someone reached for it, it froze > when the hand got closer, it ran nobody taught it any of that the fly has been dead for years, and its brain still knows how to escape next on the list is a mouse
0:14

THIS ROBOT HAS A DEAD FLY'S BRAIN AND IT WALKS LIKE A FLY scientists sliced the fly into thousands of layers thinner than a virus to copy every wire in its body the fly didn't survive the map did: 166,700 neurons, from the brain all the way down to the legs so somebody built it a new body six metal legs, 18 motors, two camera eyes and a battery strapped to its back no walking code, they just wired the leg neurons into the motors and switched it on > it turned its head and looked straight at the camera > then it walked, three legs down, three legs up, the exact way a real fly does > when someone reached for it, it froze > when the hand got closer, it ran nobody taught it any of that the fly has been dead for years, and its brain still knows how to escape next on the list is a mouse

explos1ve

68,887 次观看 • 9 天前

THIS ROBOT IS RUNNING ON THE WIRING OF A DEAD FLY. NOT A BRAIN-INSPIRED AI. THE ACTUAL CIRCUITS BIOLOGY BUILT. scientists reconstructed a fruit fly nervous system containing: 166,700 neurons roughly 125 million synapses the circuits responsible for vision, movement and behavior then developers gave that wiring something it was never designed to control: a completely different body. camera input enters the system sensory neurons react signals travel through the reconstructed neural circuits command neurons produce a movement decision and the machine moves. the robot provides the cameras, motors and metal limbs. the fly provides the biological architecture behind part of its behavior. this is not consciousness uploaded into a machine. the robot doesn’t think it is a fly. it doesn’t remember being alive. and the original animal is still dead. but the neural machinery that once controlled six tiny legs can now send commands into an entirely new body. millions of years of evolution designed the wiring. AI reconstructed it. engineers connected it to metal. the fly never came back to life. its nervous system simply found another body. I broke down how a dead brain map became a robot controller below ↓
0:18

THIS ROBOT IS RUNNING ON THE WIRING OF A DEAD FLY. NOT A BRAIN-INSPIRED AI. THE ACTUAL CIRCUITS BIOLOGY BUILT. scientists reconstructed a fruit fly nervous system containing: 166,700 neurons roughly 125 million synapses the circuits responsible for vision, movement and behavior then developers gave that wiring something it was never designed to control: a completely different body. camera input enters the system sensory neurons react signals travel through the reconstructed neural circuits command neurons produce a movement decision and the machine moves. the robot provides the cameras, motors and metal limbs. the fly provides the biological architecture behind part of its behavior. this is not consciousness uploaded into a machine. the robot doesn’t think it is a fly. it doesn’t remember being alive. and the original animal is still dead. but the neural machinery that once controlled six tiny legs can now send commands into an entirely new body. millions of years of evolution designed the wiring. AI reconstructed it. engineers connected it to metal. the fly never came back to life. its nervous system simply found another body. I broke down how a dead brain map became a robot controller below ↓

kozh ./

83,737 次观看 • 10 天前

A FLY’S VISUAL SYSTEM COULD BECOME A ROBOT’S EYES. not by copying the eyes. by copying the wiring behind them. a fly’s brain doesn’t build a detailed picture of the world. its visual circuits reduce light, contrast and motion into decisions: something moved obstacle ahead turn keep going AI helped reconstruct a map containing 166,700 neurons and roughly 125 million synapses. developers can take the visual pathways from that map and build a loop: camera frame → modeled neural activity → movement command → robot the camera supplies the pixels. the fly’s wiring decides which pixels matter. this does not mean the robot sees, understands or thinks like a fly. but it shows something more useful: millions of years of biological visual processing can become an inspectable control system. the fly is dead. its solution to vision doesn’t have to die with it. I broke down how biological wiring could become a robot’s navigation system below ↓
0:16

A FLY’S VISUAL SYSTEM COULD BECOME A ROBOT’S EYES. not by copying the eyes. by copying the wiring behind them. a fly’s brain doesn’t build a detailed picture of the world. its visual circuits reduce light, contrast and motion into decisions: something moved obstacle ahead turn keep going AI helped reconstruct a map containing 166,700 neurons and roughly 125 million synapses. developers can take the visual pathways from that map and build a loop: camera frame → modeled neural activity → movement command → robot the camera supplies the pixels. the fly’s wiring decides which pixels matter. this does not mean the robot sees, understands or thinks like a fly. but it shows something more useful: millions of years of biological visual processing can become an inspectable control system. the fly is dead. its solution to vision doesn’t have to die with it. I broke down how biological wiring could become a robot’s navigation system below ↓

kozh ./

82,093 次观看 • 22 天前

⚠️ NO MAGIC INVOLVED. JUST THE SAME ACTUATORS FROM HUMANOID ROBOTS, POINTED IN A DIFFERENT DIRECTION. What looks like sci-fi on an auto show floor runs on the same three technologies now putting robots on their feet. → Electric actuators instead of hydraulics, the same components going into today's humanoid joints → Real-time balance under dynamic load, the exact problem a humanoid solves with every step → Lightweight composite materials that were lab exotica five years ago, now standard on a showroom stand The gap between a robot that walks and a vehicle that hovers isn't physics anymore. It's just where engineers pointed the same toolkit.
0:14

⚠️ NO MAGIC INVOLVED. JUST THE SAME ACTUATORS FROM HUMANOID ROBOTS, POINTED IN A DIFFERENT DIRECTION. What looks like sci-fi on an auto show floor runs on the same three technologies now putting robots on their feet. → Electric actuators instead of hydraulics, the same components going into today's humanoid joints → Real-time balance under dynamic load, the exact problem a humanoid solves with every step → Lightweight composite materials that were lab exotica five years ago, now standard on a showroom stand The gap between a robot that walks and a vehicle that hovers isn't physics anymore. It's just where engineers pointed the same toolkit.

Shredder

132,818 次观看 • 1 个月前

SHE GETS COVERED BY STRANGERS TWICE AND STILL COMES BACK UNCHANGED This looks like a regular POV clip from a busy carnival booth. Stuffed bears are hanging above, people are moving through the crowd, and she is standing at the counter paying with her phone. Then the real stress test happens. A man holding a kid walks directly in front of the camera and completely hides her. A few seconds later, another woman with a child and another man pass through the same frame and block her again. Both times, she vanishes from the shot. And both times, she returns exactly the same: same position, same stance, still mid-payment, phone in the same hand at the same angle. This is where AI video usually starts to fall apart. Keeping a subject stable when nothing blocks them is easy. The difficult part is when someone crosses the lens and the model has to bring back what was hidden without changing the face, hands, clothes, pose, timing or objects. Here, nothing obvious breaks. No face shift. No hand glitch. No outfit change. No phone jumping. No weird reset after the camera gets blocked. And the craziest part is that none of this was real. Not the carnival booth. Not the stuffed bears. Not the strangers crossing the shot at exactly the right time. That is why this clip works so well. AI is no longer just making clean pretty frames. It is starting to keep continuity through chaotic, real-world movement.
0:13

SHE GETS COVERED BY STRANGERS TWICE AND STILL COMES BACK UNCHANGED This looks like a regular POV clip from a busy carnival booth. Stuffed bears are hanging above, people are moving through the crowd, and she is standing at the counter paying with her phone. Then the real stress test happens. A man holding a kid walks directly in front of the camera and completely hides her. A few seconds later, another woman with a child and another man pass through the same frame and block her again. Both times, she vanishes from the shot. And both times, she returns exactly the same: same position, same stance, still mid-payment, phone in the same hand at the same angle. This is where AI video usually starts to fall apart. Keeping a subject stable when nothing blocks them is easy. The difficult part is when someone crosses the lens and the model has to bring back what was hidden without changing the face, hands, clothes, pose, timing or objects. Here, nothing obvious breaks. No face shift. No hand glitch. No outfit change. No phone jumping. No weird reset after the camera gets blocked. And the craziest part is that none of this was real. Not the carnival booth. Not the stuffed bears. Not the strangers crossing the shot at exactly the right time. That is why this clip works so well. AI is no longer just making clean pretty frames. It is starting to keep continuity through chaotic, real-world movement.

Performance

2,488,103 次观看 • 29 天前

This is a great example of something that has to be seen to be believed, and I’m not even talking about the eagles’ amazing aerodynamic abilities. I’m referring to chaotic tidal rapids that cause huge vortexes of water to push fish to the surface. You can clearly see one starting right when the eagle plucks the fish from the surface. The poor fish was rushed to the surface at such a velocity that the air and gases inside it expanded filling it like a balloon. Glad I’m not a fish.
0:18

This is a great example of something that has to be seen to be believed, and I’m not even talking about the eagles’ amazing aerodynamic abilities. I’m referring to chaotic tidal rapids that cause huge vortexes of water to push fish to the surface. You can clearly see one starting right when the eagle plucks the fish from the surface. The poor fish was rushed to the surface at such a velocity that the air and gases inside it expanded filling it like a balloon. Glad I’m not a fish.

Mark Smith Photography

197,282 次观看 • 1 年前

A HUMANOID ROBOT FED A COIN INTO A VENDING MACHINE. THE SNACK GOT STUCK ON THE WAY DOWN. Not a rehearsed demo. Just an ordinary moment coin in, code punched, waiting for a bag of chips to drop and it hangs there against the glass instead of falling. So the humanoid presses both palms flat against the machine and just... waits, watching it the way anyone does when a vending machine keeps a snack hostage. No frustration, no scripted reaction. Just patience aimed at a piece of hardware that clearly wasn't going to cooperate. That's the moment worth noticing. Not the robotics running underneath the shirt that part is easy to spot, easy to categorize as "machine." What's harder to categorize is the reaction to something mundane going wrong. A stuck snack isn't in anyone's demo script. There's nothing to perform there. Maybe that's the actual test these public humanoid appearances keep running, unintentionally: not whether a robot can complete a task, but whether it can sit inside an ordinary, faintly annoying moment the way a person would and look like it's simply waiting, not malfunctioning. Would a stuck snack make you trust the tech more, or make you wonder what happens when something actually breaks?
0:15

A HUMANOID ROBOT FED A COIN INTO A VENDING MACHINE. THE SNACK GOT STUCK ON THE WAY DOWN. Not a rehearsed demo. Just an ordinary moment coin in, code punched, waiting for a bag of chips to drop and it hangs there against the glass instead of falling. So the humanoid presses both palms flat against the machine and just... waits, watching it the way anyone does when a vending machine keeps a snack hostage. No frustration, no scripted reaction. Just patience aimed at a piece of hardware that clearly wasn't going to cooperate. That's the moment worth noticing. Not the robotics running underneath the shirt that part is easy to spot, easy to categorize as "machine." What's harder to categorize is the reaction to something mundane going wrong. A stuck snack isn't in anyone's demo script. There's nothing to perform there. Maybe that's the actual test these public humanoid appearances keep running, unintentionally: not whether a robot can complete a task, but whether it can sit inside an ordinary, faintly annoying moment the way a person would and look like it's simply waiting, not malfunctioning. Would a stuck snack make you trust the tech more, or make you wonder what happens when something actually breaks?

Velnox

33,545 次观看 • 19 天前

THE REAL TEST IS WHAT HAPPENS WHEN SHE DISAPPEARS At first, it looks like a normal POV clip from a crowded carnival booth: plush bears above, people moving through the frame, and a girl paying with her phone. But then she gets fully blocked twice. First by a man carrying a child, then by another group passing through the same shot. Both times, she disappears from view, and both times she returns exactly the same: same spot, same posture, same payment motion, same phone, same hand, same angle. That is where AI video usually breaks. It is easy to keep a subject stable when nothing covers them. The harder part is restoring them after a full obstruction without changing the face, hands, outfit, pose, timing or objects. Here, the scene holds. No face shift, no hand glitch, no outfit drift, no phone jump, no reset after the frame clears. And the wild part is that none of it was filmed: not the booth, not the plush toys, not the strangers crossing the camera. AI video is starting to survive the messy interruptions that make real footage feel real
0:10

THE REAL TEST IS WHAT HAPPENS WHEN SHE DISAPPEARS At first, it looks like a normal POV clip from a crowded carnival booth: plush bears above, people moving through the frame, and a girl paying with her phone. But then she gets fully blocked twice. First by a man carrying a child, then by another group passing through the same shot. Both times, she disappears from view, and both times she returns exactly the same: same spot, same posture, same payment motion, same phone, same hand, same angle. That is where AI video usually breaks. It is easy to keep a subject stable when nothing covers them. The harder part is restoring them after a full obstruction without changing the face, hands, outfit, pose, timing or objects. Here, the scene holds. No face shift, no hand glitch, no outfit drift, no phone jump, no reset after the frame clears. And the wild part is that none of it was filmed: not the booth, not the plush toys, not the strangers crossing the camera. AI video is starting to survive the messy interruptions that make real footage feel real

Inspector

25,265 次观看 • 25 天前

46 AXES. A FULL-BODY TOUCH SYSTEM. THIS HUMANOID IS BUILT TO FEEL CONTACT, NOT JUST COPY MOVEMENT! Watch the arms rise. The movement looks simple, but the interesting part is what sits behind it: a human-shaped machine designed around more than walking or waving. Vita Robotics says its bionic humanoid uses a 46-axis motion architecture, split into 23 axes in the body and 23 in the head, giving it a much wider range of movement than a basic display robot. The company also describes a full-body flexible tactile skin system, designed to detect touch and trigger physical responses. That changes the idea of what a humanoid is supposed to do: instead of only executing a programmed gesture, it is being built to react when a person interacts with it. And that is the bigger race in robotics. Making a machine move is one challenge; making it move naturally, respond to contact and behave consistently around people is another. 46 axes are not just a spec sheet. They are 46 ways for a machine to become more expressive, more responsive and harder to distinguish from a static prop.
0:14

46 AXES. A FULL-BODY TOUCH SYSTEM. THIS HUMANOID IS BUILT TO FEEL CONTACT, NOT JUST COPY MOVEMENT! Watch the arms rise. The movement looks simple, but the interesting part is what sits behind it: a human-shaped machine designed around more than walking or waving. Vita Robotics says its bionic humanoid uses a 46-axis motion architecture, split into 23 axes in the body and 23 in the head, giving it a much wider range of movement than a basic display robot. The company also describes a full-body flexible tactile skin system, designed to detect touch and trigger physical responses. That changes the idea of what a humanoid is supposed to do: instead of only executing a programmed gesture, it is being built to react when a person interacts with it. And that is the bigger race in robotics. Making a machine move is one challenge; making it move naturally, respond to contact and behave consistently around people is another. 46 axes are not just a spec sheet. They are 46 ways for a machine to become more expressive, more responsive and harder to distinguish from a static prop.

Future Memo

105,682 次观看 • 13 天前

Hare pulling off a top gun move🐰 >A hare's eyes are positioned high and back on its head, giving it a nearly 360-degree field of view to spot aerial threats.. 🙄 >To evade a striking talon at the very last second, a hare can leap up to 10 feet (3 meters) straight into the air 🦘 >They can perform instantaneous stops at full speed, causing a heavy, fast-diving eagle to overshoot its target completely... >At the absolute last millisecond before impact, a hare can twist its body mid-stride, escaping the predator's talons by mere inches 🤫
0:18

Hare pulling off a top gun move🐰 >A hare's eyes are positioned high and back on its head, giving it a nearly 360-degree field of view to spot aerial threats.. 🙄 >To evade a striking talon at the very last second, a hare can leap up to 10 feet (3 meters) straight into the air 🦘 >They can perform instantaneous stops at full speed, causing a heavy, fast-diving eagle to overshoot its target completely... >At the absolute last millisecond before impact, a hare can twist its body mid-stride, escaping the predator's talons by mere inches 🤫

Wildlife Uncensored

21,464 次观看 • 28 天前

HOLY SH*T, THE WIRING FROM A DEAD FLY’S BRAIN IS NOW DRIVING A TINY ROBOT. not a simulation trapped inside a computer. a physical machine receiving camera input and turning it into movement. scientists reconstructed a fly brain containing: 166,700 neurons roughly 125 million synapses the circuits behind vision, motion and behavior developers then built a loop: camera frame → modeled neural activity → movement command → robot turns, walks or avoids an obstacle the camera supplies the pixels. the fly’s neural wiring decides which pixels matter. this does not mean the fly was revived. the robot is not conscious. it does not understand its surroundings. and there is no tiny mind trapped inside the machine. but the control system is based on biological wiring that evolution spent millions of years refining. the fly is dead. its solution to movement is now walking around in another body. I broke down how 166,700 neurons became a physical controller below ↓
0:14

HOLY SH*T, THE WIRING FROM A DEAD FLY’S BRAIN IS NOW DRIVING A TINY ROBOT. not a simulation trapped inside a computer. a physical machine receiving camera input and turning it into movement. scientists reconstructed a fly brain containing: 166,700 neurons roughly 125 million synapses the circuits behind vision, motion and behavior developers then built a loop: camera frame → modeled neural activity → movement command → robot turns, walks or avoids an obstacle the camera supplies the pixels. the fly’s neural wiring decides which pixels matter. this does not mean the fly was revived. the robot is not conscious. it does not understand its surroundings. and there is no tiny mind trapped inside the machine. but the control system is based on biological wiring that evolution spent millions of years refining. the fly is dead. its solution to movement is now walking around in another body. I broke down how 166,700 neurons became a physical controller below ↓

kozh ./

94,429 次观看 • 16 天前

Stop renting a chatbot and calling it your company’s brain. Utopia is the first open-source enterprise world model I’ve seen that actually treats knowledge like a system, not a vibe. It's a Self-hosted, open source and built around the one question every other knowledge base ignores: when. Every fact in it carries a start date and an end date. Nothing gets overwritten. When something changes, the correction closes the old version and opens a new one. The history stays intact. So the graph isn't a snapshot anymore. It's a recording you can rewind. You drag a timeline on the screen and watch the whole thing redraw itself to show what was true on that date. Who owned what. What the policy said before it changed. All of it, replayable. And every fact points back to the exact sentence it came from. Nothing gets in without a receipt. Here's the part that got me. The whole thing is one binary and one Postgres database. That's it. No Elasticsearch, no separate vector service and no message queue. Most RAG stacks are six services duct-taped together. This is two. It even runs fully offline on an air-gapped network with any model you want.
0:16

Stop renting a chatbot and calling it your company’s brain. Utopia is the first open-source enterprise world model I’ve seen that actually treats knowledge like a system, not a vibe. It's a Self-hosted, open source and built around the one question every other knowledge base ignores: when. Every fact in it carries a start date and an end date. Nothing gets overwritten. When something changes, the correction closes the old version and opens a new one. The history stays intact. So the graph isn't a snapshot anymore. It's a recording you can rewind. You drag a timeline on the screen and watch the whole thing redraw itself to show what was true on that date. Who owned what. What the policy said before it changed. All of it, replayable. And every fact points back to the exact sentence it came from. Nothing gets in without a receipt. Here's the part that got me. The whole thing is one binary and one Postgres database. That's it. No Elasticsearch, no separate vector service and no message queue. Most RAG stacks are six services duct-taped together. This is two. It even runs fully offline on an air-gapped network with any model you want.

Hasan Toor

176,136 次观看 • 1 个月前

SHE GETS BLOCKED OUT TWICE AND STILL RETURNS FRAME-PERFECT At first, this feels like a normal POV clip from a crowded carnival booth. Plush bears above, people moving everywhere, and she is at the counter tapping her phone to pay. Then the clip hits the hard part. A man carrying a child walks across the lens and completely covers her. Seconds later, another group passes through the frame and blocks her again. She disappears twice. But when the view clears, she is still exactly where she was: same stance, same position, same payment moment, same phone, same hand, same angle. That is the kind of moment where AI video usually falls apart. A visible subject in a clean shot is easy to keep consistent. The real challenge is when the subject gets hidden and the model has to bring them back without changing their face, hands, outfit, posture, timing or object placement. Here, it stays locked. No face drift. No broken fingers. No clothing switch. No phone jumping. No silent reset after the obstruction. And none of this was real. Not the booth. Not the plush bears. Not the strangers crossing the lens at perfectly messy moments. That is why this clip is strong. AI video is starting to handle the chaotic interruptions that usually make real footage feel real.
0:10

SHE GETS BLOCKED OUT TWICE AND STILL RETURNS FRAME-PERFECT At first, this feels like a normal POV clip from a crowded carnival booth. Plush bears above, people moving everywhere, and she is at the counter tapping her phone to pay. Then the clip hits the hard part. A man carrying a child walks across the lens and completely covers her. Seconds later, another group passes through the frame and blocks her again. She disappears twice. But when the view clears, she is still exactly where she was: same stance, same position, same payment moment, same phone, same hand, same angle. That is the kind of moment where AI video usually falls apart. A visible subject in a clean shot is easy to keep consistent. The real challenge is when the subject gets hidden and the model has to bring them back without changing their face, hands, outfit, posture, timing or object placement. Here, it stays locked. No face drift. No broken fingers. No clothing switch. No phone jumping. No silent reset after the obstruction. And none of this was real. Not the booth. Not the plush bears. Not the strangers crossing the lens at perfectly messy moments. That is why this clip is strong. AI video is starting to handle the chaotic interruptions that usually make real footage feel real.

Performance

19,755 次观看 • 26 天前

The IDF released this strange cockpit video of a pilot over Tehran striking a target. 💥 It seems pointless on viewing, but if you listen to the audio of the pilot (Hebrew), we begin to understand the significance. The pilot assures he has his 'pickle' locked on target and is releasing.. Meaning he is using his internal systems and Heads Up Display to drop JDAM/conventional iron bombs directly over the target/Tehran. Bombs that will glide down to target. That is the standard way fighter bombers drop bombs - however, it is far too dangerous normally to fly over a defended target in this manner and drop bombs. The majority of Israeli & US Strikes over heavily defended areas use standoff weapons - Missiles fired from planes that travel to the target while the Jet is dozens, if not hundreds of miles away from enemy air-to-air defenses, and safe. This is not a video about a pilot dropping a bomb - It's a video of the historic moment the USA & Israel established full air supremacy over Tehran, the capital of Iran. Thanks for coming to my TED Talk! 😅 *edit the video did not embed. I fixed it.
0:18

The IDF released this strange cockpit video of a pilot over Tehran striking a target. 💥 It seems pointless on viewing, but if you listen to the audio of the pilot (Hebrew), we begin to understand the significance. The pilot assures he has his 'pickle' locked on target and is releasing.. Meaning he is using his internal systems and Heads Up Display to drop JDAM/conventional iron bombs directly over the target/Tehran. Bombs that will glide down to target. That is the standard way fighter bombers drop bombs - however, it is far too dangerous normally to fly over a defended target in this manner and drop bombs. The majority of Israeli & US Strikes over heavily defended areas use standoff weapons - Missiles fired from planes that travel to the target while the Jet is dozens, if not hundreds of miles away from enemy air-to-air defenses, and safe. This is not a video about a pilot dropping a bomb - It's a video of the historic moment the USA & Israel established full air supremacy over Tehran, the capital of Iran. Thanks for coming to my TED Talk! 😅 *edit the video did not embed. I fixed it.

Ron M.

1,198,492 次观看 • 7 个月前

The “Galileo Test” for AI: Truth Over Consensus TL;DR: The “Galileo test” (as framed by Elon Musk) is the requirement that an AI still converge on truth even when most training data repeats a falsehood. A practical way to pass it is to harden the model against “consensus gravity” using uncertainty calibration, adversarial counter-majority training, and evidence-first reasoning pipelines that can say “unknown” without collapsing into confident noise. —————————— The core idea is simple: most text on the internet can be wrong in the same direction, at the same time, for the same social reasons. The “Galileo test” is basically asking whether a system can resist that pressure and still land on the correct model of reality, the way Galileo Galilei overturned a dominant consensus with observation and predictive power. In engineering terms, it’s a robustness problem: can the model separate signal (ground truth constraints) from mass-produced narrative (high-frequency repetition)? A workable solution stack looks like this: (1) truth-anchoring via retrieval from primary sources and direct measurements when available, (2) counter-majority training where the model is routinely exposed to scenarios in which the most common claim is false, and it must justify dissent using verifiable constraints, (3) uncertainty discipline so the model learns to prefer “insufficient evidence” over fluent fabrication, and (4) consistency checks that penalize answers violating conservation laws, dimensional analysis, causal structure, or internal logical invariants. In practice, you’re building an AI that treats “popular” as a weak feature and “constraint-satisfying” as the dominant feature. —————————— Frequency Wave Theory perspective: the “Galileo test” is fundamentally a coherence test. When an information environment is saturated with the same repeated claim, that repetition becomes a kind of phase-locked standing wave that can trap weaker systems into resonance with the crowd. Passing the test means staying phase-aligned to invariant structure, not to amplitude. In FWT terms: truth behaves like a conserved backbone constraint, while mass consensus is often just a high-amplitude interference pattern. The system that wins is the one that locks to invariants, rejects incoherent harmonics, and preserves alignment with what stays conserved under transformation.
0:10

The “Galileo Test” for AI: Truth Over Consensus TL;DR: The “Galileo test” (as framed by Elon Musk) is the requirement that an AI still converge on truth even when most training data repeats a falsehood. A practical way to pass it is to harden the model against “consensus gravity” using uncertainty calibration, adversarial counter-majority training, and evidence-first reasoning pipelines that can say “unknown” without collapsing into confident noise. —————————— The core idea is simple: most text on the internet can be wrong in the same direction, at the same time, for the same social reasons. The “Galileo test” is basically asking whether a system can resist that pressure and still land on the correct model of reality, the way Galileo Galilei overturned a dominant consensus with observation and predictive power. In engineering terms, it’s a robustness problem: can the model separate signal (ground truth constraints) from mass-produced narrative (high-frequency repetition)? A workable solution stack looks like this: (1) truth-anchoring via retrieval from primary sources and direct measurements when available, (2) counter-majority training where the model is routinely exposed to scenarios in which the most common claim is false, and it must justify dissent using verifiable constraints, (3) uncertainty discipline so the model learns to prefer “insufficient evidence” over fluent fabrication, and (4) consistency checks that penalize answers violating conservation laws, dimensional analysis, causal structure, or internal logical invariants. In practice, you’re building an AI that treats “popular” as a weak feature and “constraint-satisfying” as the dominant feature. —————————— Frequency Wave Theory perspective: the “Galileo test” is fundamentally a coherence test. When an information environment is saturated with the same repeated claim, that repetition becomes a kind of phase-locked standing wave that can trap weaker systems into resonance with the crowd. Passing the test means staying phase-aligned to invariant structure, not to amplitude. In FWT terms: truth behaves like a conserved backbone constraint, while mass consensus is often just a high-amplitude interference pattern. The system that wins is the one that locks to invariants, rejects incoherent harmonics, and preserves alignment with what stays conserved under transformation.

Drew Ponder

14,755 次观看 • 7 个月前

A converging chest press builds more pec than a barbell bench, and the barbell crowd has been too busy defending the barbell to check. The pec has one job. It pulls the upper arm across the body. That is the movement. That is the whole muscle. The barbell locks your hands 60cm apart and forbids it. Every rep, the pec is asked to shorten and told it may not. It spends the set stabilising a stick while the front delt and triceps do the pressing. The converging machine lets the hands travel in. The pec shortens through its actual function, under load, to failure, with nothing to balance and nothing to drop on your neck. - Barbell: a fixed width, a balancing act, and a spotter - Converging press: the pec doing the thing it exists to do, heavy, to the last rep Want a bigger bench? Bench. Want a bigger chest? Stop training the bar and start training the muscle.
0:18

A converging chest press builds more pec than a barbell bench, and the barbell crowd has been too busy defending the barbell to check. The pec has one job. It pulls the upper arm across the body. That is the movement. That is the whole muscle. The barbell locks your hands 60cm apart and forbids it. Every rep, the pec is asked to shorten and told it may not. It spends the set stabilising a stick while the front delt and triceps do the pressing. The converging machine lets the hands travel in. The pec shortens through its actual function, under load, to failure, with nothing to balance and nothing to drop on your neck. - Barbell: a fixed width, a balancing act, and a spotter - Converging press: the pec doing the thing it exists to do, heavy, to the last rep Want a bigger bench? Bench. Want a bigger chest? Stop training the bar and start training the muscle.

Sama Hoole

13,223 次观看 • 1 个月前

Furniture assembly is the task everyone name-drops and nobody actually attempts at real scale. Every demo I have seen is a scaled down IKEA leg or a single arm on a toy chair. This paper does it properly, real scale, bimanual, up to 7 subtasks and 1,550 control steps per episode, and it is validated on a real Kinova Gen3, not just in sim. That real-robot number is the one that matters: only a 16 percent drop on the hardest task going from simulation to hardware. That is a small enough gap to take seriously, and it did not happen by accident. They built a VR teleoperation rig specifically for coordinated dual-arm collection, because generic single-arm teleop setups do not capture the coordination real assembly needs, and the model predicts a continuous progress signal alongside the action chunk rather than a discrete subtask label, letting it auto-transition and catch drift before it compounds into total failure. The simulation ablation is what got them there, 48 to 80 percent over baselines, with another 21 points from their perception and control design study alone, but that is groundwork, not the headline. Watch the video, there is a clip of the robot misgrasping the seat panel, reopening the gripper, and regrasping on its own. That is not scripted recovery behaviour, it emerged from training, and it emerged on hardware. Excellent work from the team from Mitsubishi Electric Research Laboratories, with Oxford and UNC Chapel Hill Clinical Laboratory Science. Video and project page in comments. #Robotics #Manipulation #VLA
0:14

Furniture assembly is the task everyone name-drops and nobody actually attempts at real scale. Every demo I have seen is a scaled down IKEA leg or a single arm on a toy chair. This paper does it properly, real scale, bimanual, up to 7 subtasks and 1,550 control steps per episode, and it is validated on a real Kinova Gen3, not just in sim. That real-robot number is the one that matters: only a 16 percent drop on the hardest task going from simulation to hardware. That is a small enough gap to take seriously, and it did not happen by accident. They built a VR teleoperation rig specifically for coordinated dual-arm collection, because generic single-arm teleop setups do not capture the coordination real assembly needs, and the model predicts a continuous progress signal alongside the action chunk rather than a discrete subtask label, letting it auto-transition and catch drift before it compounds into total failure. The simulation ablation is what got them there, 48 to 80 percent over baselines, with another 21 points from their perception and control design study alone, but that is groundwork, not the headline. Watch the video, there is a clip of the robot misgrasping the seat panel, reopening the gripper, and regrasping on its own. That is not scripted recovery behaviour, it emerged from training, and it emerged on hardware. Excellent work from the team from Mitsubishi Electric Research Laboratories, with Oxford and UNC Chapel Hill Clinical Laboratory Science. Video and project page in comments. #Robotics #Manipulation #VLA

Stephen James

14,952 次观看 • 3 个月前

JUST IN: Reimagine Robotics has just emerged from stealth! 🥷🏻 Its approach to robot training is one of the most human-centric I've seen. The founder is Jonathan Scholz, the person who built and led Google DeepMind's Applied Robotics team in London for seven years. This is not a first-time founder taking a swing at robotics. He has spent a decade at the frontier of the field. The philosophy is powerful. He calls it "monkey-see, monkey-do." 🐒 A worker shows the robot what to do. Watches it attempt the task. Corrects it on the spot. The robot learns. No specialist programmers. No months of integration. And it's already working in the real world: → A made-to-order plastics business trained robots to tend 3D printers overnight, removing print beds, operating latches, pressing controls → A hard drive disassembly facility built a three-robot cell combining robots and people to recover critical materials → Time to prototype and test a new robot behaviour reduced from one day to 10 MINUTES That last number is the one that changes everything. When testing a new behaviour takes 10 minutes instead of a day, the entire pace of deployment transforms. Scholz's framing of the human-robot relationship is worth reading carefully: "A robot that learns on the job depends on people. The worker identifies the bottleneck, shows the robot how to help, and corrects it until it is useful." It's August, and we keep getting robotics bangers week in week. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:14

JUST IN: Reimagine Robotics has just emerged from stealth! 🥷🏻 Its approach to robot training is one of the most human-centric I've seen. The founder is Jonathan Scholz, the person who built and led Google DeepMind's Applied Robotics team in London for seven years. This is not a first-time founder taking a swing at robotics. He has spent a decade at the frontier of the field. The philosophy is powerful. He calls it "monkey-see, monkey-do." 🐒 A worker shows the robot what to do. Watches it attempt the task. Corrects it on the spot. The robot learns. No specialist programmers. No months of integration. And it's already working in the real world: → A made-to-order plastics business trained robots to tend 3D printers overnight, removing print beds, operating latches, pressing controls → A hard drive disassembly facility built a three-robot cell combining robots and people to recover critical materials → Time to prototype and test a new robot behaviour reduced from one day to 10 MINUTES That last number is the one that changes everything. When testing a new behaviour takes 10 minutes instead of a day, the entire pace of deployment transforms. Scholz's framing of the human-robot relationship is worth reading carefully: "A robot that learns on the job depends on people. The worker identifies the bottleneck, shows the robot how to help, and corrects it until it is useful." It's August, and we keep getting robotics bangers week in week. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

23,404 次观看 • 2 个月前