Loading video...

Video Failed to Load

Go Home

CHINA JUST SOLVED THE PROBLEM THAT'S BEEN BREAKING ROBOT AI FOR A DECADE. and the fix wasn't a smarter model. for years, every robot AI failure got the same diagnosis. the model isn't smart enough. so everyone scaled intelligence. bigger models. more parameters. better reasoning. AGIBOT asked a different...

heyshrutimishra's profile picture

Shruti

178,476 subscribers

18,622 views • 3 months ago •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

2. NVIDIA is building the GPT of humanoid robots. They just launched Isaac GR00T N1.5 - a foundation model for general purpose robotics. Here’s how it works: → A human demos the task once → Cosmos (their physics AI model) generates 1,000s of variations → Omniverse simulates the motions in high fidelity → The robot trains entirely in simulation → Then fine-tunes itself in the real world Robots can now learn general skills across tasks, tools, even body types with just one human demo. AI isn’t just thinking in text anymore. It’s perceiving. Reasoning. Moving. Physical AI is here and it’s training itself.
2:21

2. NVIDIA is building the GPT of humanoid robots. They just launched Isaac GR00T N1.5 - a foundation model for general purpose robotics. Here’s how it works: → A human demos the task once → Cosmos (their physics AI model) generates 1,000s of variations → Omniverse simulates the motions in high fidelity → The robot trains entirely in simulation → Then fine-tunes itself in the real world Robots can now learn general skills across tasks, tools, even body types with just one human demo. AI isn’t just thinking in text anymore. It’s perceiving. Reasoning. Moving. Physical AI is here and it’s training itself.

Shruti

70,359 views • 1 year ago

Robot trained entirely in simulation! 🤯 Sudo AI just released #sudo R1. This could be the moment the data bottleneck in robotics finally breaks, a major step forward for getting simulation-trained robots to actually work in the real world. #sudo R1 is a manipulation foundation model that reliably picks objects it has never seen: glass, fabric, metal, irregular shapes. All with zero real-world training data. No tele-op. No demos. No manual labeling. Pure simulation. And picking isn't just any skill, it's the foundation. Almost everything we want robots to do starts with being able to reliably pick something up. Data collection is the bottleneck holding the whole field back. Every new task means hours of tele-op and armies of labelers. If simulation alone can actually deliver production-grade picking, that bottleneck starts to disappear. Really exciting to see! If the results hold, the path to general-purpose robots just got a whole lot shorter. 👇 Read more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:59

Robot trained entirely in simulation! 🤯 Sudo AI just released #sudo R1. This could be the moment the data bottleneck in robotics finally breaks, a major step forward for getting simulation-trained robots to actually work in the real world. #sudo R1 is a manipulation foundation model that reliably picks objects it has never seen: glass, fabric, metal, irregular shapes. All with zero real-world training data. No tele-op. No demos. No manual labeling. Pure simulation. And picking isn't just any skill, it's the foundation. Almost everything we want robots to do starts with being able to reliably pick something up. Data collection is the bottleneck holding the whole field back. Every new task means hours of tele-op and armies of labelers. If simulation alone can actually deliver production-grade picking, that bottleneck starts to disappear. Really exciting to see! If the results hold, the path to general-purpose robots just got a whole lot shorter. 👇 Read more here: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

25,718 views • 2 months ago

"A parcel with snacks has been delivered for Flexion. Retrieve it using the stairs and come up using the elevator. Then unpack it and place the items into the empty drawer on the shelf in the snack area." One instruction. No human operator. Everything that follows is autonomous. Today we're introducing Reflect v1.0, our robotics intelligence platform for long-horizon work. From a single natural-language command, the robot understands the task, navigates a multi-floor building, calls elevators, handles doors, uses tools to unpack a box, and puts the items away. The biggest shift in v1.0 is that we use reinforcement learning across every layer, from low-level control to high-level reasoning. Long-horizon autonomy is unforgiving. The robot must recover on its own when things don't go to plan because in the real world, they never do. Combining reasoning, perception, physical execution and runtime robustness into a single mission-capable system is the foundation required to solve humanoid autonomy. Our team is just getting started. #HumanoidRobots #Flexion
4:17

"A parcel with snacks has been delivered for Flexion. Retrieve it using the stairs and come up using the elevator. Then unpack it and place the items into the empty drawer on the shelf in the snack area." One instruction. No human operator. Everything that follows is autonomous. Today we're introducing Reflect v1.0, our robotics intelligence platform for long-horizon work. From a single natural-language command, the robot understands the task, navigates a multi-floor building, calls elevators, handles doors, uses tools to unpack a box, and puts the items away. The biggest shift in v1.0 is that we use reinforcement learning across every layer, from low-level control to high-level reasoning. Long-horizon autonomy is unforgiving. The robot must recover on its own when things don't go to plan because in the real world, they never do. Combining reasoning, perception, physical execution and runtime robustness into a single mission-capable system is the foundation required to solve humanoid autonomy. Our team is just getting started. #HumanoidRobots #Flexion

Flexion

134,315 views • 17 days ago

We trained a robot dog to balance and walk on top of a yoga ball purely in simulation, and then transfer zero-shot to the real world. No fine-tuning. Just works. I’m excited to announce DrEureka, an LLM agent that writes code to train robot skills in simulation, and writes more code to bridge the difficult simulation-reality gap. It fully automates the pipeline from new skill learning to real-world deployment. The Yoga ball task is particularly hard because it is not possible to accurately simulate the bouncy ball surface. Yet DrEureka has no trouble searching over a vast space of sim-to-real configurations, and enables the dog to steer the ball on various terrains, even walking sideways! Traditionally, the sim-to-real transfer is achieved by domain randomization, a tedious process that requires expert human roboticists to stare at every parameter and adjust by hand. Frontier LLMs like GPT-4 have tons of built-in physical intuition for friction, damping, stiffness, gravity, etc. We are (mildly) surprised to find that DrEureka can tune these parameters competently and explain its reasoning well. DrEureka builds on our prior work Eureka, the algorithm that teaches a 5-finger robot hand to do pen spinning. It takes one step further on our quest to automate the entire robot learning pipeline by an AI agent system. One model that outputs strings will supervise another model that outputs torque control. We open-source everything! Welcome you all to check out the paper, more videos, and try the codebase today: Code:
0:24

We trained a robot dog to balance and walk on top of a yoga ball purely in simulation, and then transfer zero-shot to the real world. No fine-tuning. Just works. I’m excited to announce DrEureka, an LLM agent that writes code to train robot skills in simulation, and writes more code to bridge the difficult simulation-reality gap. It fully automates the pipeline from new skill learning to real-world deployment. The Yoga ball task is particularly hard because it is not possible to accurately simulate the bouncy ball surface. Yet DrEureka has no trouble searching over a vast space of sim-to-real configurations, and enables the dog to steer the ball on various terrains, even walking sideways! Traditionally, the sim-to-real transfer is achieved by domain randomization, a tedious process that requires expert human roboticists to stare at every parameter and adjust by hand. Frontier LLMs like GPT-4 have tons of built-in physical intuition for friction, damping, stiffness, gravity, etc. We are (mildly) surprised to find that DrEureka can tune these parameters competently and explain its reasoning well. DrEureka builds on our prior work Eureka, the algorithm that teaches a 5-finger robot hand to do pen spinning. It takes one step further on our quest to automate the entire robot learning pipeline by an AI agent system. One model that outputs strings will supervise another model that outputs torque control. We open-source everything! Welcome you all to check out the paper, more videos, and try the codebase today: Code:

Jim Fan

908,690 views • 2 years ago

Foundation models are enough to solve robotics! Unfortunately, this is not true. We keep hearing that Vision-Language-Action (VLA) models struggle because of the gap between static training and the dynamic real world. A German startup (Sereact) just released a solution that bridges this gap perfectly. They are introducing a new paradigm called Interactive RL Policy Patching. It's a distributed framework that allows robots to self-learn from human corrections without needing full retraining. When a robot fails, a human operator provides a brief "patch" or demonstration. The system then uses online off-policy reinforcement learning to update the behavior instantly. This is powered by a massive foundation model trained on hundreds of millions of interactions from over 100 deployed robot stations. The best part is the distributed parameter synchronization... When one robot learns a fix, the update is published fleet-wide... meaning the entire swarm gets smarter from a single human intervention. They are already proving this on complex manipulation tasks like shoe unboxing and screw sorting, drastically reducing the data needed to handle edge cases. Real-world environments are unforgiving, and I love seeing systems that can actually adapt on the fly! 📍 More info:
0:38

Foundation models are enough to solve robotics! Unfortunately, this is not true. We keep hearing that Vision-Language-Action (VLA) models struggle because of the gap between static training and the dynamic real world. A German startup (Sereact) just released a solution that bridges this gap perfectly. They are introducing a new paradigm called Interactive RL Policy Patching. It's a distributed framework that allows robots to self-learn from human corrections without needing full retraining. When a robot fails, a human operator provides a brief "patch" or demonstration. The system then uses online off-policy reinforcement learning to update the behavior instantly. This is powered by a massive foundation model trained on hundreds of millions of interactions from over 100 deployed robot stations. The best part is the distributed parameter synchronization... When one robot learns a fix, the update is published fleet-wide... meaning the entire swarm gets smarter from a single human intervention. They are already proving this on complex manipulation tasks like shoe unboxing and screw sorting, drastically reducing the data needed to handle edge cases. Real-world environments are unforgiving, and I love seeing systems that can actually adapt on the fly! 📍 More info:

Ilir Aliu

18,210 views • 5 months ago

A humanoid robot just unloaded a dishwasher for 67 hours straight. By end of 2026, the same robot will perform surgery. That's not a prediction from a sci-fi writer. That's Brett Adcock. CEO of Figure AI. Said it on camera this month. Here's what makes this insane. Two years ago Figure's robot put a Keurig cup in a coffee maker. That was the breakthrough moment. The whole team celebrated. A coffee cup. Today that same robot runs 67 consecutive hours in a BMW factory. Grabbing packages. Finding barcodes. Placing them. One operation every second. Just a neural network that learned by watching. And Brett just said the hands that do that the same fingers, the same sensors, the same neural net are already capable of surgical precision from a hardware standpoint. "From a hardware perspective in 2026 we'll be able to do what surgeons can do." The AI just has to catch up to the hands. Two years ago: a coffee cup. This year: a factory. Next: the operating table. The robot didn't change, the learning did and it's not stopping.
1:43:47

A humanoid robot just unloaded a dishwasher for 67 hours straight. By end of 2026, the same robot will perform surgery. That's not a prediction from a sci-fi writer. That's Brett Adcock. CEO of Figure AI. Said it on camera this month. Here's what makes this insane. Two years ago Figure's robot put a Keurig cup in a coffee maker. That was the breakthrough moment. The whole team celebrated. A coffee cup. Today that same robot runs 67 consecutive hours in a BMW factory. Grabbing packages. Finding barcodes. Placing them. One operation every second. Just a neural network that learned by watching. And Brett just said the hands that do that the same fingers, the same sensors, the same neural net are already capable of surgical precision from a hardware standpoint. "From a hardware perspective in 2026 we'll be able to do what surgeons can do." The AI just has to catch up to the hands. Two years ago: a coffee cup. This year: a factory. Next: the operating table. The robot didn't change, the learning did and it's not stopping.

Shruti

46,761 views • 2 months ago

📢 Announcing one of the most exciting works from us this year on **scalable robot policy evaluation through real-to-sim transfer**, moving toward a scalable evaluation engine with structured world models that capture the appearance, geometry, and dynamics of environments involving deformable objects. 🤖 Evaluation remains one of the biggest bottlenecks in building general-purpose robots. Today, robots are still evaluated only in the real world, which is **orders of magnitude slower** than the development of language agents. We propose a new framework where simulation performance **strongly correlates** with the real world (r > 0.9), even for deformable objects. The key difference from existing work lies in the correlation between simulation and reality: if a robot model performs better in the digital world, does it also perform better in the real world? This question has long made people hesitant about simulation-based evaluation — especially for deformable objects. We are changing that. Our pipeline achieves effective real-to-sim transfer, establishing **state-of-the-art correlation** between simulation and reality for deformable object manipulation. It provides a **scalable and reproducible evaluation engine** for robot learning. 🌐
3:04

📢 Announcing one of the most exciting works from us this year on **scalable robot policy evaluation through real-to-sim transfer**, moving toward a scalable evaluation engine with structured world models that capture the appearance, geometry, and dynamics of environments involving deformable objects. 🤖 Evaluation remains one of the biggest bottlenecks in building general-purpose robots. Today, robots are still evaluated only in the real world, which is **orders of magnitude slower** than the development of language agents. We propose a new framework where simulation performance **strongly correlates** with the real world (r > 0.9), even for deformable objects. The key difference from existing work lies in the correlation between simulation and reality: if a robot model performs better in the digital world, does it also perform better in the real world? This question has long made people hesitant about simulation-based evaluation — especially for deformable objects. We are changing that. Our pipeline achieves effective real-to-sim transfer, establishing **state-of-the-art correlation** between simulation and reality for deformable object manipulation. It provides a **scalable and reproducible evaluation engine** for robot learning. 🌐

Yunzhu Li

39,900 views • 8 months ago

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:
0:29

Tired of teleoperation? One human video → 1,000s of robot demos. (📍GitHub ) Scaling Robot Data Without Dynamics Simulation or Robot Hardware Real2Render2Real (R2R2R) is a new way to scale robot data without physics simulation or hardware. You take a phone scan + a single monocular human demo. It tracks the motion, renders photorealistic scenes, and generates diverse, robot-agnostic trajectories ready for training. > No teleop, no sim, no robot, just a phone and a video > Train VLA models and diffusion policies directly on the output > Supports multiple robot embodiments with kinematic consistency > 1000s of demos in 1/27 the time of real-world collection Thank you, Max Fu, for sharing!! Project: Paper: Code coming soon: It shows that with the right pipeline, you can scale robot learning data without touching a robot. One of the most interesting directions in scalable robotics today. —— Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

42,804 views • 5 months ago

The teams shipping AI agents right now are bleeding money on the dumbest possible expense: teaching a 400B-parameter model to read a file name. Every time an AI agent needs to "see" something today, it routes an image through a frontier model. OCR, object detection, checking if a button exists on screen. You're paying GPT-4o or Claude pricing for tasks that require perception, not reasoning. One agent workflow processing a few thousand screenshots per day can burn through more on vision calls than on the actual thinking. Perceptron's Isaac is 2B parameters. Built by the team that created Meta's Chameleon multimodal models. On perceptive benchmarks, it matches or beats models 50x its size. The VQA, OCR, and object detection scores are competitive with models running on infrastructure that costs orders of magnitude more. The MCP wrapper is the distribution play. One install command and every Claude Code agent can offload vision tasks to a model that runs on a single consumer GPU. The agent keeps its reasoning in the frontier model and routes perception to a specialist. That split is how you get vision-heavy agent workflows from "technically possible but expensive" to "cheap enough to run on everything." This is the same pattern that won in every other compute-intensive stack. General-purpose handles orchestration. Specialists handle the heavy lifting. Graphics went through it. Audio went through it. Video encoding went through it. Vision in AI agents is next. The teams building agents that see 10,000 images a day will care about this before anyone else does.
0:38

The teams shipping AI agents right now are bleeding money on the dumbest possible expense: teaching a 400B-parameter model to read a file name. Every time an AI agent needs to "see" something today, it routes an image through a frontier model. OCR, object detection, checking if a button exists on screen. You're paying GPT-4o or Claude pricing for tasks that require perception, not reasoning. One agent workflow processing a few thousand screenshots per day can burn through more on vision calls than on the actual thinking. Perceptron's Isaac is 2B parameters. Built by the team that created Meta's Chameleon multimodal models. On perceptive benchmarks, it matches or beats models 50x its size. The VQA, OCR, and object detection scores are competitive with models running on infrastructure that costs orders of magnitude more. The MCP wrapper is the distribution play. One install command and every Claude Code agent can offload vision tasks to a model that runs on a single consumer GPU. The agent keeps its reasoning in the frontier model and routes perception to a specialist. That split is how you get vision-heavy agent workflows from "technically possible but expensive" to "cheap enough to run on everything." This is the same pattern that won in every other compute-intensive stack. General-purpose handles orchestration. Specialists handle the heavy lifting. Graphics went through it. Audio went through it. Video encoding went through it. Vision in AI agents is next. The teams building agents that see 10,000 images a day will care about this before anyone else does.

Aakash Gupta

55,978 views • 3 months ago

AI for robotics has some monumental challenges, like high dimensionality and reasoning with the physical world. Ashok Elluswamy, AI lead at Tesla, thinks the hardest problem is Evaluation. - Loss isn't a perfect indicator of Policy Neural Network's quality. - Testing in the real world for all the long-tail failure cases is not possible. To solve the evaluation challenge, Tesla has developed a World Simulator Model. It is trained on state and action pairs, so for a given current camera feed and current action, the model generates the next frame in the cameras. Once the world simulation is good enough, it can be hooked up to the policy network for a closed-loop simulation. In this setup, the model generates the next video frames, and based on those, the policy network comes up with the next action, which feeds back into the world simulation; this continuous loop creates a coherent simulation. The world simulation neural network is not just applicable to self-driving but also to humanoids, because Tesla trains it on common data. The examples show Optimus navigating or manipulating inside the world simulation, with all the pixels being generated by the world simulation model. The robotic agents can be tested in an accurately reproduced edge case and in other potential variations of the same edge case.
1:29

AI for robotics has some monumental challenges, like high dimensionality and reasoning with the physical world. Ashok Elluswamy, AI lead at Tesla, thinks the hardest problem is Evaluation. - Loss isn't a perfect indicator of Policy Neural Network's quality. - Testing in the real world for all the long-tail failure cases is not possible. To solve the evaluation challenge, Tesla has developed a World Simulator Model. It is trained on state and action pairs, so for a given current camera feed and current action, the model generates the next frame in the cameras. Once the world simulation is good enough, it can be hooked up to the policy network for a closed-loop simulation. In this setup, the model generates the next video frames, and based on those, the policy network comes up with the next action, which feeds back into the world simulation; this continuous loop creates a coherent simulation. The world simulation neural network is not just applicable to self-driving but also to humanoids, because Tesla trains it on common data. The examples show Optimus navigating or manipulating inside the world simulation, with all the pixels being generated by the world simulation model. The robotic agents can be tested in an accurately reproduced edge case and in other potential variations of the same edge case.

The Humanoid Hub

16,591 views • 5 months ago

"Everyone's talking about continual learning. That's entirely where this space is going to go." The Applied Compute platform is architected around that premise: build memory and intuition from fragmented data across your entire org, train reasoning models directly on top of it, and close the loop. A model is just one piece. An agent is where it runs, what tools it has, how permissions and auth are handled, how humans guide and instruct it, and the observability around it all. Every interaction should be treated as a training signal so the system can compound over time. Thanks for having us TBPN
1:46

"Everyone's talking about continual learning. That's entirely where this space is going to go." The Applied Compute platform is architected around that premise: build memory and intuition from fragmented data across your entire org, train reasoning models directly on top of it, and close the loop. A model is just one piece. An agent is where it runs, what tools it has, how permissions and auth are handled, how humans guide and instruct it, and the observability around it all. Every interaction should be treated as a training signal so the system can compound over time. Thanks for having us TBPN

Applied Compute

64,246 views • 4 months ago

Better data for robot hands! 🤲🏼 Embodied reasoning is the bottleneck now. It’s the data we are missing. Specifically, understanding what the robot's hands are doing throughout a task. Perceptron AI new Egocentric offering tracks both hands through an entire video instead of guessing from sampled frames. That matters because a lot happens between frames. Hand positions shift. Contacts change. The robot adjusts grip. You miss all that if you're sampling. 0.280 semantic end-to-end F1 score on WGO-Bench vs 0.158 from the best pipelines built on Gemini Robotics-ER 1.6. Training a robot to manipulate objects requires understanding every moment of hand-object interaction. If your training data has gaps: missing frames, missed contacts, unclear hand positions, the policy learns from incomplete information. You end up with robots that work in demos but fail in real tasks. Continuous hand tracking gives you the full picture. The policy learns from cleaner, more complete data. That compounds into better real-world performance. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:25

Better data for robot hands! 🤲🏼 Embodied reasoning is the bottleneck now. It’s the data we are missing. Specifically, understanding what the robot's hands are doing throughout a task. Perceptron AI new Egocentric offering tracks both hands through an entire video instead of guessing from sampled frames. That matters because a lot happens between frames. Hand positions shift. Contacts change. The robot adjusts grip. You miss all that if you're sampling. 0.280 semantic end-to-end F1 score on WGO-Bench vs 0.158 from the best pipelines built on Gemini Robotics-ER 1.6. Training a robot to manipulate objects requires understanding every moment of hand-object interaction. If your training data has gaps: missing frames, missed contacts, unclear hand positions, the policy learns from incomplete information. You end up with robots that work in demos but fail in real tasks. Continuous hand tracking gives you the full picture. The policy learns from cleaner, more complete data. That compounds into better real-world performance. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

17,926 views • 7 days ago

watch this anon. i gave NVIDIA's biggest model ever a single task. 100 minutes and 440,000 tokens later, it had rendered nothing. not one important thing on the screen. this is Nemotron 3 Ultra. 550 billion parameters, a hybrid Mamba Transformer MoE, the largest model NVIDIA has ever shipped, and they built it specifically for long-running agentic coding. so i handed it exactly that: build a 3D scene from a spec, multiple files, iterate until the tests pass. the same task a frontier model one shotted in minutes. i genuinely wanted to be impressed. it ran for an hour and forty. burned through 440,000 tokens. wrote every file, passed its own tests, and proudly printed "task complete."the browser was blank. the 3D scene never rendered. not once. and the long horizon agentic behavior was genuinely good. it stayed on task the whole hour and forty, wrote real multi-file code, drove its own tools without derailing. it just couldn't turn any of that into something that actually runs. here's the part that gets me. it's a text model, it cannot see its own output. so it sat there looping on a broken vision tool, trying to "look" at the page, hitting error after error, never once reasoning its way out. it declared victory on an empty screen because it had no way to know the screen was empty. to be fair, i genuinely don't know what quant the NIM was serving, so maybe some of that's on the serving, not the model. but the biggest model NVIDIA has ever made, on the exact task it was designed for, couldn't tell it had built nothing in 100 minutes. same task on a local model, below thread👇.
4:07

watch this anon. i gave NVIDIA's biggest model ever a single task. 100 minutes and 440,000 tokens later, it had rendered nothing. not one important thing on the screen. this is Nemotron 3 Ultra. 550 billion parameters, a hybrid Mamba Transformer MoE, the largest model NVIDIA has ever shipped, and they built it specifically for long-running agentic coding. so i handed it exactly that: build a 3D scene from a spec, multiple files, iterate until the tests pass. the same task a frontier model one shotted in minutes. i genuinely wanted to be impressed. it ran for an hour and forty. burned through 440,000 tokens. wrote every file, passed its own tests, and proudly printed "task complete."the browser was blank. the 3D scene never rendered. not once. and the long horizon agentic behavior was genuinely good. it stayed on task the whole hour and forty, wrote real multi-file code, drove its own tools without derailing. it just couldn't turn any of that into something that actually runs. here's the part that gets me. it's a text model, it cannot see its own output. so it sat there looping on a broken vision tool, trying to "look" at the page, hitting error after error, never once reasoning its way out. it declared victory on an empty screen because it had no way to know the screen was empty. to be fair, i genuinely don't know what quant the NIM was serving, so maybe some of that's on the serving, not the model. but the biggest model NVIDIA has ever made, on the exact task it was designed for, couldn't tell it had built nothing in 100 minutes. same task on a local model, below thread👇.

Sudo su

32,589 views • 17 days ago

JUST IN: Figure has gone from producing 1 humanoid robot per day to 1 per hour, a 24x throughput improvement in under 120 days. This data is from their manufacturing facility BotQ update (link below) Here's what's happening inside BotQ: → Over 350 Figure 03 robots delivered and counting → End-of-line first-pass yield now above 80% and improving weekly → Battery line at 99.3% first-pass yield with 500+ battery packs shipped → Over 9,000 actuators produced across 10+ distinct SKUs → Every robot put through 80+ functional verification tests before sign-off Their Helix AI model can now SEE the environment it's moving through, not just feel it. Stairs, ramps, uneven terrain, navigated zero-shot, trained entirely in simulation, deployed directly to hardware with no real-world fine-tuning. Brett Adcock has been clear about the plan: every robot that rolls off the line is a data-collection engine. The bigger the fleet, the faster the AI improves. The faster the AI improves, the more deployable the robot becomes. This is the flywheel. And it is now spinning. The humanoid robot race just got very serious. VERY INTERESTING. Alright fam, what are your thoughts on Figure? 👀 Link: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:44

JUST IN: Figure has gone from producing 1 humanoid robot per day to 1 per hour, a 24x throughput improvement in under 120 days. This data is from their manufacturing facility BotQ update (link below) Here's what's happening inside BotQ: → Over 350 Figure 03 robots delivered and counting → End-of-line first-pass yield now above 80% and improving weekly → Battery line at 99.3% first-pass yield with 500+ battery packs shipped → Over 9,000 actuators produced across 10+ distinct SKUs → Every robot put through 80+ functional verification tests before sign-off Their Helix AI model can now SEE the environment it's moving through, not just feel it. Stairs, ramps, uneven terrain, navigated zero-shot, trained entirely in simulation, deployed directly to hardware with no real-world fine-tuning. Brett Adcock has been clear about the plan: every robot that rolls off the line is a data-collection engine. The bigger the fleet, the faster the AI improves. The faster the AI improves, the more deployable the robot becomes. This is the flywheel. And it is now spinning. The humanoid robot race just got very serious. VERY INTERESTING. Alright fam, what are your thoughts on Figure? 👀 Link: ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

38,067 views • 2 months ago

a team of researchers just proved you don't need a bigger model, you need a smarter plan researchers from Tsinghua and South China University of Technology built a framework called Atomic Task Graph. it turned 7B-8B open-source models into GPT-4 competitors on complex agent benchmarks, beating it on two out of three. no fine-tuning. no extra training. zero parameter updates. current AI agents plan in a straight line. step 1, step 2, step 3. when step 4 fails, the whole chain breaks. and the longer the chain gets, the more the model hallucinates because it's reasoning over a ballooning text history. here's how it works. 1. instead of a linear chain, ATG breaks any complex task into a directed graph where subtask inputs and outputs are explicitly mapped 2. it recursively decomposes each subtask until every node is one atomic tool call 3. independent branches run in parallel instead of waiting in line 4. before anything executes, a lightweight "thought experiment" simulates the plan internally to catch bad dependencies and missing steps early 5. when something breaks at runtime, ATG traces the failure to the exact subgraph that caused it and repairs only that piece. validated work stays frozen. the old way meant a failure at step 5 forced a full replan from scratch. hallucinated actions piled up the longer the task ran. ReAct hit a 43% hallucination rate on household tasks. ATG on an 8B Llama model scored 63.65 on ALFWorld. GPT-4 with ReAct scored 41.24 on the same benchmark. hallucinated actions dropped to 12%. those numbers happened because someone stopped throwing compute at the problem and started thinking about how work gets organized. that's the part that gets me. the industry is spending billions on scale. this team spent time on architecture. and the architecture won.
0:23

a team of researchers just proved you don't need a bigger model, you need a smarter plan researchers from Tsinghua and South China University of Technology built a framework called Atomic Task Graph. it turned 7B-8B open-source models into GPT-4 competitors on complex agent benchmarks, beating it on two out of three. no fine-tuning. no extra training. zero parameter updates. current AI agents plan in a straight line. step 1, step 2, step 3. when step 4 fails, the whole chain breaks. and the longer the chain gets, the more the model hallucinates because it's reasoning over a ballooning text history. here's how it works. 1. instead of a linear chain, ATG breaks any complex task into a directed graph where subtask inputs and outputs are explicitly mapped 2. it recursively decomposes each subtask until every node is one atomic tool call 3. independent branches run in parallel instead of waiting in line 4. before anything executes, a lightweight "thought experiment" simulates the plan internally to catch bad dependencies and missing steps early 5. when something breaks at runtime, ATG traces the failure to the exact subgraph that caused it and repairs only that piece. validated work stays frozen. the old way meant a failure at step 5 forced a full replan from scratch. hallucinated actions piled up the longer the task ran. ReAct hit a 43% hallucination rate on household tasks. ATG on an 8B Llama model scored 63.65 on ALFWorld. GPT-4 with ReAct scored 41.24 on the same benchmark. hallucinated actions dropped to 12%. those numbers happened because someone stopped throwing compute at the problem and started thinking about how work gets organized. that's the part that gets me. the industry is spending billions on scale. this team spent time on architecture. and the architecture won.

Alex Veremeyenko

171,258 views • 5 days ago

Boston Dynamics just gave its robot dog a brain that reasons about the physical world. Google DeepMind's Gemini Robotics model is now running inside Spot, the four-legged robot already deployed at thousands of facilities worldwide. The upgrade centers on something called embodied reasoning. Basically, instead of following rigid instructions, Spot now interprets what it sees and makes decisions on its own. One standout example: Spot can now look at a pressure gauge, zoom in on the needle, calculate the intervals between tick marks, and deliver an accurate reading. All autonomously. The model also introduces: > Multi-view reasoning, combining feeds from multiple cameras to understand its environment even when objects are blocked from one angle > Built-in success detection, so Spot knows when it's failed and decides whether to retry or move on That last part is a key requirement for true autonomy. Most companies building AI-powered robots are still stuck in the demo phase. Boston Dynamics is one of the few with actual paying customers... which means thousands of Spots collecting real-world data that feeds back into making the AI smarter. That flywheel is hard to compete with.
1:39

Boston Dynamics just gave its robot dog a brain that reasons about the physical world. Google DeepMind's Gemini Robotics model is now running inside Spot, the four-legged robot already deployed at thousands of facilities worldwide. The upgrade centers on something called embodied reasoning. Basically, instead of following rigid instructions, Spot now interprets what it sees and makes decisions on its own. One standout example: Spot can now look at a pressure gauge, zoom in on the needle, calculate the intervals between tick marks, and deliver an accurate reading. All autonomously. The model also introduces: > Multi-view reasoning, combining feeds from multiple cameras to understand its environment even when objects are blocked from one angle > Built-in success detection, so Spot knows when it's failed and decides whether to retry or move on That last part is a key requirement for true autonomy. Most companies building AI-powered robots are still stuck in the demo phase. Boston Dynamics is one of the few with actual paying customers... which means thousands of Spots collecting real-world data that feeds back into making the AI smarter. That flywheel is hard to compete with.

Rowan Cheung

54,376 views • 2 months ago

Dr. Fei-Fei Li just called out the biggest blind spot in the entire AI industry. We have been building half of human intelligence. And calling it the finish line. Li: “If you look at human intelligence, it pretty much boils down to two buckets.” The first bucket is language. Symbolic reasoning. Communication. The ability to think in words and abstractions. That’s what every major AI lab has spent the last decade building. The second bucket is the one the industry has almost entirely ignored. Li: “We call that in AI spatial intelligence.” How humans and animals perceive, navigate, and interact with the three-dimensional physical world. How we reach for objects. How we move through space. How we build and manipulate physical reality. From painting masterpieces to constructing the pyramids, non-verbal spatial intelligence is what actually shapes the world. Language describes reality. Spatial intelligence acts on it. And the gap between those two things is the gap between a chatbot and a robot. Li: “When this technology is ready, the robotic revolution is gonna start. We’re already seeing that trend.” Every robot is a moving agent. Every moving agent requires spatial intelligence to function in the real world. The humanoid robots being deployed in factories right now are hitting the ceiling of what language models alone can power. Spatial intelligence is the unlock. But Li didn’t stop at robotics. Li: “From a geopolitics point of view, this is part of the technology that goes straight into weapons.” Autonomous drone swarms. Battlefield navigation. Physical target acquisition without human oversight. Every military application of AI that operates in the real world runs on spatial intelligence. The nation that masters the transition from static text to dynamic three-dimensional perception doesn’t just win the software race. It commands the physical battlefield. The AI arms race just broke out of the data center. It’s operating in three dimensions now.
3:08

Dr. Fei-Fei Li just called out the biggest blind spot in the entire AI industry. We have been building half of human intelligence. And calling it the finish line. Li: “If you look at human intelligence, it pretty much boils down to two buckets.” The first bucket is language. Symbolic reasoning. Communication. The ability to think in words and abstractions. That’s what every major AI lab has spent the last decade building. The second bucket is the one the industry has almost entirely ignored. Li: “We call that in AI spatial intelligence.” How humans and animals perceive, navigate, and interact with the three-dimensional physical world. How we reach for objects. How we move through space. How we build and manipulate physical reality. From painting masterpieces to constructing the pyramids, non-verbal spatial intelligence is what actually shapes the world. Language describes reality. Spatial intelligence acts on it. And the gap between those two things is the gap between a chatbot and a robot. Li: “When this technology is ready, the robotic revolution is gonna start. We’re already seeing that trend.” Every robot is a moving agent. Every moving agent requires spatial intelligence to function in the real world. The humanoid robots being deployed in factories right now are hitting the ceiling of what language models alone can power. Spatial intelligence is the unlock. But Li didn’t stop at robotics. Li: “From a geopolitics point of view, this is part of the technology that goes straight into weapons.” Autonomous drone swarms. Battlefield navigation. Physical target acquisition without human oversight. Every military application of AI that operates in the real world runs on spatial intelligence. The nation that masters the transition from static text to dynamic three-dimensional perception doesn’t just win the software race. It commands the physical battlefield. The AI arms race just broke out of the data center. It’s operating in three dimensions now.

Dustin

122,680 views • 4 months ago

Today, we give robots a /skills library that self-evolves and compounds indefinitely! Introducing ASPIRE: a robot solving its 100th task is no longer as clueless as solving its first. Coding agents observe multimodal sensory traces from simulation and real robots, launch an evolutionary search over control programs, and distill the best know-how into an ever-expanding library. ASPIRE is a new type of continual learning: "training" is skill refinement instead of gradient descent. "Trained model" is a repo of sensorimotor skills instead of floating weights. “Distributed training” is a panel of agents each practicing a different skill instead of sharded minibatches. Here's the beauty: ASPIRE gives the tired terms "sim2real transfer" and "cross-embodiment transfer" a whole new meaning. Bridging the sim-to-real gap is notoriously brutal. An end-to-end policy has to swallow both the visual shift (sim looks toyish next to a real camera) and the subtle contact physics it never quite gets right. ASPIRE sidesteps the mess, because it doesn't ship pixels or weights across the gap, but ships the know-how. The robot still has to practice in the real world, not zero-shot, but it gets there way faster because it isn't rediscovering the strategy from scratch. Same for going single-arm to bimanual hardware, which usually requires new data and retraining from zero. ASPIRE achieves up to ~10x cut in "transfer learning” tokens (yes, tokens are the new unit of *training* compute ;) Check out our gallery of 150+ tasks and 90+ skills the robots taught themselves, all on the website! Kind of wild that we can ship the "learned weights" as an HTML page rather than a GGUF. We'll open-source the full stack so your own robot library starts compounding from ours! Deep dive in thread:
0:40

Today, we give robots a /skills library that self-evolves and compounds indefinitely! Introducing ASPIRE: a robot solving its 100th task is no longer as clueless as solving its first. Coding agents observe multimodal sensory traces from simulation and real robots, launch an evolutionary search over control programs, and distill the best know-how into an ever-expanding library. ASPIRE is a new type of continual learning: "training" is skill refinement instead of gradient descent. "Trained model" is a repo of sensorimotor skills instead of floating weights. “Distributed training” is a panel of agents each practicing a different skill instead of sharded minibatches. Here's the beauty: ASPIRE gives the tired terms "sim2real transfer" and "cross-embodiment transfer" a whole new meaning. Bridging the sim-to-real gap is notoriously brutal. An end-to-end policy has to swallow both the visual shift (sim looks toyish next to a real camera) and the subtle contact physics it never quite gets right. ASPIRE sidesteps the mess, because it doesn't ship pixels or weights across the gap, but ships the know-how. The robot still has to practice in the real world, not zero-shot, but it gets there way faster because it isn't rediscovering the strategy from scratch. Same for going single-arm to bimanual hardware, which usually requires new data and retraining from zero. ASPIRE achieves up to ~10x cut in "transfer learning” tokens (yes, tokens are the new unit of *training* compute ;) Check out our gallery of 150+ tasks and 90+ skills the robots taught themselves, all on the website! Kind of wild that we can ship the "learned weights" as an HTML page rather than a GGUF. We'll open-source the full stack so your own robot library starts compounding from ours! Deep dive in thread:

Jim Fan

198,213 views • 16 days ago

That's sick! 🤯 Genesis AI simulates robots playing yo-yo! 🪀 Genesis AI just open-sourced Genesis World 1.0, and it might be one of the most important infrastructure releases in robotics this year. Robotics is still bottlenecked by the 1× speed of the physical world. Every model needs to be tested on real hardware, slowly, expensively, with limited coverage. Genesis World 1.0 from Genesis AI flips that equation: One hour in reality becomes 100 days in simulation. That turns a wall-clock bottleneck into a compute problem. And compute problems are solvable. The technical stack they rebuilt from scratch is serious: → GPU-accelerated cross-platform compiler via Quadrants, 10x faster launch time and up to 4.6x runtime vs the initial Genesis release → Penetration-free multi-physics contact solvers, the thing that makes simulation actually trustworthy → Unified rigid AND deformable physics in a single engine → Nyx, a high-performance path-traced rendering engine purpose-built for physical AI The sim-to-real gap has historically been the graveyard of robotics research. Policies that work beautifully in simulation fall apart on real hardware. Genesis World 1.0 is a direct attack on that problem. And it's fully open-source. The companies that master simulation infrastructure will train better robots faster than anyone else. Find it here: Genesis World 1.0: Quadrants: Nyx: Theophile Gervet, Zhou Xian congrats! 👏🏼 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →
0:46

That's sick! 🤯 Genesis AI simulates robots playing yo-yo! 🪀 Genesis AI just open-sourced Genesis World 1.0, and it might be one of the most important infrastructure releases in robotics this year. Robotics is still bottlenecked by the 1× speed of the physical world. Every model needs to be tested on real hardware, slowly, expensively, with limited coverage. Genesis World 1.0 from Genesis AI flips that equation: One hour in reality becomes 100 days in simulation. That turns a wall-clock bottleneck into a compute problem. And compute problems are solvable. The technical stack they rebuilt from scratch is serious: → GPU-accelerated cross-platform compiler via Quadrants, 10x faster launch time and up to 4.6x runtime vs the initial Genesis release → Penetration-free multi-physics contact solvers, the thing that makes simulation actually trustworthy → Unified rigid AND deformable physics in a single engine → Nyx, a high-performance path-traced rendering engine purpose-built for physical AI The sim-to-real gap has historically been the graveyard of robotics research. Policies that work beautifully in simulation fall apart on real hardware. Genesis World 1.0 is a direct attack on that problem. And it's fully open-source. The companies that master simulation infrastructure will train better robots faster than anyone else. Find it here: Genesis World 1.0: Quadrants: Nyx: Theophile Gervet, Zhou Xian congrats! 👏🏼 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

36,767 views • 1 month ago