Video wird geladen...

Video konnte nicht geladen werden

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

Dr Fei-Fei-Li explains with a simple example how everyday household chores are so extremely difficult for Robots. "If you tell a robot to open the top drawer and watch out for the vase, this is actually a really hard task for robots." because the robot must ground language into... the real world. Words like "top", "drawer", and "vase" are abstract. The system has to map them to 3D locations, objects, and relations in a noisy scene. This requires robust perception, object recognition, and spatial reasoning under uncertainty. The robot also lacks human commonsense. "Watch out" implies predicting consequences, estimating clearances, and understanding that vases are fragile. Encoding such priors, like how heavy a drawer is or how a vase might tip, is very complex and difficult without rich world knowledge. Learning the behavior from rewards is tough. The success signal is very sparse here, so naive exploration almost never stumbles on a full success sequence. This makes policy learning sample inefficient and brittle, especially when the environment changes between training and deployment. A sparse reward situation is when the agent only gets a success signal at the very end, and gets little or no feedback along the way. If a robot must open a drawer without hitting a vase, it might get reward only if the drawer ends up open and the vase is intact. Every partial try before that looks the same to the learner, reward equals 0. --- From "DSAI by Dr. Osbert Tay" YT channelshow more

Rohan Paul

154,001 subscribers

342,627 Aufrufe • vor 9 Monaten •via X (Twitter)

Bildung Gesundheit & Wellness Wissenschaft & Technologie

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

$We’re excited to launch Instance – a success detector for robot rollouts! Describe your task and drop in a dataset, and it judges success/fail on each rollout with detailed subtask captions– it’s more accurate than Claude Opus 4.8, at a fraction of the latency. This is just the first step toward our larger goal: an autonomous evaluation rig for robot learning. Today, evaluating a robot policy means humans have to watch the robot roll out, mark success, and reset the scene– we're automating that, starting with the success judge.$

We’re excited to launch Instance – a success detector for robot rollouts! Describe your task and drop in a dataset, and it judges success/fail on each rollout with detailed subtask captions– it’s more accurate than Claude Opus 4.8, at a fraction of the latency. This is just the first step toward our larger goal: an autonomous evaluation rig for robot learning. Today, evaluating a robot policy means humans have to watch the robot roll out, mark success, and reset the scene– we're automating that, starting with the success judge.

Lucy Cai

73,929 Aufrufe • vor 1 Monat

a robot can recognize every object in a room and still not know how to open a single drawer knowing there's a cabinet is scene understanding. knowing which handle to grab, which direction to pull, and that it's a drawer not a door is affordance understanding almost no dataset has it in 3D SceneFun3D: 14,800+ functional elements across 710 laser-scanned indoor scenes. every handle, knob, button, and switch labeled with its affordance, motion axis, motion type, and a natural language task description each scene's 3D point cloud with its iPad video recordings in fiftyone. the affordance annotations live in 3D and project into the video frames check it out here

a robot can recognize every object in a room and still not know how to open a single drawer knowing there's a cabinet is scene understanding. knowing which handle to grab, which direction to pull, and that it's a drawer not a door is affordance understanding almost no dataset has it in 3D SceneFun3D: 14,800+ functional elements across 710 laser-scanned indoor scenes. every handle, knob, button, and switch labeled with its affordance, motion axis, motion type, and a natural language task description each scene's 3D point cloud with its iPad video recordings in fiftyone. the affordance annotations live in 3D and project into the video frames check it out here

harpreet

18,800 Aufrufe • vor 2 Monaten

Robot policies must be both reliable and highly capable to be useful; the best way to achieve this level of performance is with reinforcement learning. However, for reinforcement learning you are usually stuck between two difficult options: reinforcement in the real world is often risky and expensive, while reinforcement learning in a traditional simulator takes a lot of engineering work and has a persistent sim-to-real gap. What if instead you could train your robot purely in a world model? RISE by Jiazhi Yang et al. uses a compositional world model to predict the future and evaluate progress. This allows for a self-improving pipeline, which learns a world model from real data and then learns how the robot should perform different tasks. This pipeline results in a data-driven way to improve policy performance from real data but without real-world reinforcement learning. Watch Episode #86 of RoboPapers, with Chris Paxton and Jiafei Duan, to learn more!

Robot policies must be both reliable and highly capable to be useful; the best way to achieve this level of performance is with reinforcement learning. However, for reinforcement learning you are usually stuck between two difficult options: reinforcement in the real world is often risky and expensive, while reinforcement learning in a traditional simulator takes a lot of engineering work and has a persistent sim-to-real gap. What if instead you could train your robot purely in a world model? RISE by Jiazhi Yang et al. uses a compositional world model to predict the future and evaluate progress. This allows for a self-improving pipeline, which learns a world model from real data and then learns how the robot should perform different tasks. This pipeline results in a data-driven way to improve policy performance from real data but without real-world reinforcement learning. Watch Episode #86 of RoboPapers, with Chris Paxton and Jiafei Duan, to learn more!

RoboPapers

38,334 Aufrufe • vor 2 Monaten

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:

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 Aufrufe • vor 7 Monaten

Open-source robot arm meets hand tracking [📍GitHub below] It is designed with an industrial mindset but built as a 3D-printed desktop system. PAROL6 paired with a LEAP Motion controller is a nice example of how accessible robot teleoperation has become. • Hand motion is streamed to the robot at 100 Hz via UDP • A pneumatic gripper is controlled by simple fist open and close gestures • The entire robot stack is open source, from mechanics to control software Combine that with low-latency hand tracking and you get a very practical platform for learning manipulation, teleoperation, and human-robot interfaces. This kind of setup is great for experimentation, teleop, data collection, and teaching robots by demonstration All without proprietary hardware or locked software. Credit to SourceRobotics 📍Code: —— Weekly robotics and AI insights. Subscribe free:

Open-source robot arm meets hand tracking [📍GitHub below] It is designed with an industrial mindset but built as a 3D-printed desktop system. PAROL6 paired with a LEAP Motion controller is a nice example of how accessible robot teleoperation has become. • Hand motion is streamed to the robot at 100 Hz via UDP • A pneumatic gripper is controlled by simple fist open and close gestures • The entire robot stack is open source, from mechanics to control software Combine that with low-latency hand tracking and you get a very practical platform for learning manipulation, teleoperation, and human-robot interfaces. This kind of setup is great for experimentation, teleop, data collection, and teaching robots by demonstration All without proprietary hardware or locked software. Credit to SourceRobotics 📍Code: —— Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

32,025 Aufrufe • vor 7 Monaten

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.

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

122,781 Aufrufe • vor 6 Monaten

Rockford, Illinois nurse showing that the blood shortage is so bad they only have half of a drawer left of O-Negative blood (universal donor) O-Negative is used very commonly in hospitals. They only have a few bags left “So to put into perspective how high the need for blood is right now, take a look at our drawers here. This is where we store our O negative blood, which is the universal blood donor. This whole drawer is empty and only have this much O-negative in our other drawer. So O-negative is used really commonly in the hospitals and we really need these to be stocked. This just goes to show you how high the need really is right now.”

Rockford, Illinois nurse showing that the blood shortage is so bad they only have half of a drawer left of O-Negative blood (universal donor) O-Negative is used very commonly in hospitals. They only have a few bags left “So to put into perspective how high the need for blood is right now, take a look at our drawers here. This is where we store our O negative blood, which is the universal blood donor. This whole drawer is empty and only have this much O-negative in our other drawer. So O-negative is used really commonly in the hospitals and we really need these to be stocked. This just goes to show you how high the need really is right now.”

Wall Street Apes

443,453 Aufrufe • vor 1 Jahr

"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

"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

141,192 Aufrufe • vor 2 Monaten

In order for robots to be deployed in the real world, performing tasks of real value, they must be reliable. Unfortunately, even more, most robotic demos work maybe 70-80% of the time at best. The way to get better reliability is to do real-world reinforcement learning: having the robot teach itself how to perform the task up to a high level of success. The key to doing this is to start with a core of expert human data, use that to train a policy then iteratively improve it, until finally finishing with on-policy reinforcement learning. Kun Lei talks through a unified framework for imitation and reinforcement learning based on PPO, which enables this improvement process. In this episode, Kun Lei explains the theory behind his reinforcement learning method and how it allowed his robot to run in a shopping mall juicing oranges for seven hours at a time, among experiments on a wide variety of tasks and embodiments. Watch episode 58 of RoboPapers now, hosted by Michael Cho - Rbt/Acc and Chris Paxton!

In order for robots to be deployed in the real world, performing tasks of real value, they must be reliable. Unfortunately, even more, most robotic demos work maybe 70-80% of the time at best. The way to get better reliability is to do real-world reinforcement learning: having the robot teach itself how to perform the task up to a high level of success. The key to doing this is to start with a core of expert human data, use that to train a policy then iteratively improve it, until finally finishing with on-policy reinforcement learning. Kun Lei talks through a unified framework for imitation and reinforcement learning based on PPO, which enables this improvement process. In this episode, Kun Lei explains the theory behind his reinforcement learning method and how it allowed his robot to run in a shopping mall juicing oranges for seven hours at a time, among experiments on a wide variety of tasks and embodiments. Watch episode 58 of RoboPapers now, hosted by Michael Cho - Rbt/Acc and Chris Paxton!

RoboPapers

18,813 Aufrufe • vor 7 Monaten

15–30 minutes of real-world robot data. That's now enough to go from sim-to-real failure to working robot. Let’s see… You train a robot in simulation. You deploy it in the real world. It fails. The physics don't match. So you try to fine-tune it with real data, but… you never have enough real data, and the fine-tuning breaks everything the simulation taught it. SimDist fixes this with one key decision: don't transfer the policy. Transfer the world model. Keep the reward and value knowledge from simulation frozen. Only update the part that's actually wrong, how the robot predicts physics. Now the robot doesn't have to relearn the entire task in the real world. It already knows what success looks like. It just needs to correct its understanding of how the real world moves. The part that makes this work: they also trained on failures and recoveries; not just perfect demonstrations. Without that, the planner finds the gaps and exploits them. With it, the robot can tell a good future from a bad one. That's all it needs. Results on peg insertion, table leg assembly, locomotion on slippery and uneven surfaces. Tasks that require precision, force, and quick reaction. Thanks for sharing, Tyler Westenbroek ([Tyler Westenbroek]. Interactive visualization + paper: ——- Weekly robotics and AI insights. Subscribe free:

15–30 minutes of real-world robot data. That's now enough to go from sim-to-real failure to working robot. Let’s see… You train a robot in simulation. You deploy it in the real world. It fails. The physics don't match. So you try to fine-tune it with real data, but… you never have enough real data, and the fine-tuning breaks everything the simulation taught it. SimDist fixes this with one key decision: don't transfer the policy. Transfer the world model. Keep the reward and value knowledge from simulation frozen. Only update the part that's actually wrong, how the robot predicts physics. Now the robot doesn't have to relearn the entire task in the real world. It already knows what success looks like. It just needs to correct its understanding of how the real world moves. The part that makes this work: they also trained on failures and recoveries; not just perfect demonstrations. Without that, the planner finds the gaps and exploits them. With it, the robot can tell a good future from a bad one. That's all it needs. Results on peg insertion, table leg assembly, locomotion on slippery and uneven surfaces. Tasks that require precision, force, and quick reaction. Thanks for sharing, Tyler Westenbroek ([Tyler Westenbroek]. Interactive visualization + paper: ——- Weekly robotics and AI insights. Subscribe free:

Ilir Aliu

14,222 Aufrufe • vor 3 Monaten

Just a little taste of the Robohacks hackathon happening now at Y Combinator in San Francisco. This is only two of the teams. The first is a team building a robot for an AirBnb on Mars. The second built a system that 3D prints new tools for the robot to do new work on Mars. Both demonstrate why I love builders so much. Dreaming of a new world and learning together how to build for a new robot none of them have used before.

Just a little taste of the Robohacks hackathon happening now at Y Combinator in San Francisco. This is only two of the teams. The first is a team building a robot for an AirBnb on Mars. The second built a system that 3D prints new tools for the robot to do new work on Mars. Both demonstrate why I love builders so much. Dreaming of a new world and learning together how to build for a new robot none of them have used before.

Robert Scoble

14,399 Aufrufe • vor 4 Monaten

$I made Physical AutoResearch sound simple (conceptually), but it took a village to pull off and lots of design thinking into the robot /loopcraft. The hardest part is everything we need to setup *before* pressing Enter. Here's a behind-the-scene tour: 1. Safety harness Letting 8 robots run unattended overnight means safety has to be more than a hint in the system prompt. ENPIRE hardwires it in 2 layers: (1) hard kinematic limit that trips an immediate task failure and auto-resets as soon as a robot leaves its safety envelope, and (2) a torque-limited compliant gripper so a bad contact or misaligned insertion ends in a safe stall, instead of crushing the robot or the object at hand. We make safety more conservative than usual so humans can sleep tight. In reality, we still need a few human operators to watch over the "robots of loving grace". 2. Definition of /done An agent that can edit its own reward will game it for sure. ENPIRE fixes the goalposts before the fleet can move them. Here's the recipe: Collect a few minutes of success & failure demos -> Ask agent to write code using computer vision tools to classify success and measure against groundtruth -> Agent hill-climbs on classifier until reliably good -> This classifier becomes the real-time reward function that directly computes on sensor streams -> *Freeze* the reward function before AutoResearch. It's sacred, enshrined in a Gym env that no one can touch. 3. System telemetry design Robot-seconds is by far the scarcest resource, followed by GPU-seconds, and finally tokens. We instrument all three and surface them to ENPIRE for live resource awareness rather than letting it hill-climb in a vacuum. We define: - Mean Robot Utilization ("MRU"): the fraction of wall-clock time when the robot is actively executing an experiment. Otherwise the hardware is sitting idle and waiting for the next code commit. - Mean Token Utilization ("MTU"): tokens consumed per minute, our proxy for how hard the agent is actually thinking. A low MTU means the agent is stalled, waiting on a robot rollout to finish instead of doing research. - GPU utilization: fraction of wall-clock time when GPU is active. ... and evaluate on two budget-to-outcome metrics: 1. Tokens-to-Success: token budget the fleet burns to complete /goal. 2. Time-to-Success: wall-clock time to /goal$

I made Physical AutoResearch sound simple (conceptually), but it took a village to pull off and lots of design thinking into the robot /loopcraft. The hardest part is everything we need to setup before pressing Enter. Here's a behind-the-scene tour: 1. Safety harness Letting 8 robots run unattended overnight means safety has to be more than a hint in the system prompt. ENPIRE hardwires it in 2 layers: (1) hard kinematic limit that trips an immediate task failure and auto-resets as soon as a robot leaves its safety envelope, and (2) a torque-limited compliant gripper so a bad contact or misaligned insertion ends in a safe stall, instead of crushing the robot or the object at hand. We make safety more conservative than usual so humans can sleep tight. In reality, we still need a few human operators to watch over the "robots of loving grace". 2. Definition of /done An agent that can edit its own reward will game it for sure. ENPIRE fixes the goalposts before the fleet can move them. Here's the recipe: Collect a few minutes of success & failure demos -> Ask agent to write code using computer vision tools to classify success and measure against groundtruth -> Agent hill-climbs on classifier until reliably good -> This classifier becomes the real-time reward function that directly computes on sensor streams -> Freeze the reward function before AutoResearch. It's sacred, enshrined in a Gym env that no one can touch. 3. System telemetry design Robot-seconds is by far the scarcest resource, followed by GPU-seconds, and finally tokens. We instrument all three and surface them to ENPIRE for live resource awareness rather than letting it hill-climb in a vacuum. We define: - Mean Robot Utilization ("MRU"): the fraction of wall-clock time when the robot is actively executing an experiment. Otherwise the hardware is sitting idle and waiting for the next code commit. - Mean Token Utilization ("MTU"): tokens consumed per minute, our proxy for how hard the agent is actually thinking. A low MTU means the agent is stalled, waiting on a robot rollout to finish instead of doing research. - GPU utilization: fraction of wall-clock time when GPU is active. ... and evaluate on two budget-to-outcome metrics: 1. Tokens-to-Success: token budget the fleet burns to complete /goal. 2. Time-to-Success: wall-clock time to /goal

Jim Fan

107,457 Aufrufe • vor 2 Monaten

Let's reverse engineer Disney's adorable, lifelike robot! I couldn't find a whitepaper, but this is how I think it's trained: 1. The emotional behaviors are curated by Disney animation artists, keyframe by keyframe. But it cannot be "rendered" directly on the robot because it doesn't take into account the complex real-world physics. 2. Reinforcement learning (RL) is a great tool for training low-level robot controllers. RL needs a reward function to optimize, and it's typically a task reward (e.g. walk in a straight line as fast as possible). The problem is that RL doesn't know what counts as "natural behavior", and often produces weird-looking body postures that somehow still maximize the reward. This is a human alignment problem just like ChatGPT. 3. Enters Adversarial Motion Prior (AMP): a technique that learns the human preference by training a classifier on what we consider "emotional & cute". In GAN literature, this is called a discriminator. Disney artists are good at creating such a dataset. You can then add AMP as an auxiliary reward in simulation to nudge the robot towards desired behaviors. AMP was developed by Peng et al. 2021 and Escontrela et al. 2022. 4. Add lots of data augmentation to make the controller robust to physical disturbances. In RL, it's called "domain randomization". This is a very powerful technique that bridges the gap between simulator and reality. Previously, OpenAI used domain randomization to train a 5-finger robot hand to manipulate a Rubik's Cube: IEEE news article gave hints about the pipeline: Finally, praying for world peace 🙏. I hope robotics like this will bring more joy to the world.

Let's reverse engineer Disney's adorable, lifelike robot! I couldn't find a whitepaper, but this is how I think it's trained: 1. The emotional behaviors are curated by Disney animation artists, keyframe by keyframe. But it cannot be "rendered" directly on the robot because it doesn't take into account the complex real-world physics. 2. Reinforcement learning (RL) is a great tool for training low-level robot controllers. RL needs a reward function to optimize, and it's typically a task reward (e.g. walk in a straight line as fast as possible). The problem is that RL doesn't know what counts as "natural behavior", and often produces weird-looking body postures that somehow still maximize the reward. This is a human alignment problem just like ChatGPT. 3. Enters Adversarial Motion Prior (AMP): a technique that learns the human preference by training a classifier on what we consider "emotional & cute". In GAN literature, this is called a discriminator. Disney artists are good at creating such a dataset. You can then add AMP as an auxiliary reward in simulation to nudge the robot towards desired behaviors. AMP was developed by Peng et al. 2021 and Escontrela et al. 2022. 4. Add lots of data augmentation to make the controller robust to physical disturbances. In RL, it's called "domain randomization". This is a very powerful technique that bridges the gap between simulator and reality. Previously, OpenAI used domain randomization to train a 5-finger robot hand to manipulate a Rubik's Cube: IEEE news article gave hints about the pipeline: Finally, praying for world peace 🙏. I hope robotics like this will bring more joy to the world.

Jim Fan

314,694 Aufrufe • vor 2 Jahren

These guys trained a robot with water bottles, and the robot learned to pick up other objects it had never practiced on. They did this using reinforcement learning: 1. The robot had to pick a bottle out of a messy tote 2. Then it had to hand the bottle to a person 3. Only water bottles. Nothing else. After they finished training, they tested the robot on different objects: • On bottles, the success went from 80% to 98% • On snack cans, the robot's throughput rose 40% • On soft bags, the robot's performance improved by 13% even though they're deformable and much harder to handle than rigid objects. Snack cans and soft bags have different shapes and weights, and the robot was never trained on them. This was just one of three use cases they tested. Every hour we spend teaching a robot costs a ton of money. If they only get better at handling the one object they practice on, reinforcement learning will never be worth doing. But if the gains carry over, we could focus on teaching the hardest possible tasks and let the robot apply that knowledge elsewhere. Here is the full write-up by Humanoid on the infrastructure they used and what they learned: #ad

These guys trained a robot with water bottles, and the robot learned to pick up other objects it had never practiced on. They did this using reinforcement learning: 1. The robot had to pick a bottle out of a messy tote 2. Then it had to hand the bottle to a person 3. Only water bottles. Nothing else. After they finished training, they tested the robot on different objects: • On bottles, the success went from 80% to 98% • On snack cans, the robot's throughput rose 40% • On soft bags, the robot's performance improved by 13% even though they're deformable and much harder to handle than rigid objects. Snack cans and soft bags have different shapes and weights, and the robot was never trained on them. This was just one of three use cases they tested. Every hour we spend teaching a robot costs a ton of money. If they only get better at handling the one object they practice on, reinforcement learning will never be worth doing. But if the gains carry over, we could focus on teaching the hardest possible tasks and let the robot apply that knowledge elsewhere. Here is the full write-up by Humanoid on the infrastructure they used and what they learned: #ad

Santiago

13,952 Aufrufe • vor 1 Monat

This work makes a humanoid robot do simple parkour moves by looking with a depth camera and choosing the right move on the fly. The big deal is that it turns lots of small human moves into long, real-time robot behavior, without hand-coding every transition or retraining for each new course. A humanoid robot is usually good at steady walking, but it often fails when it has to do fast moves like jumping up, vaulting, or rolling, and then keep going to the next obstacle. The hard part is that you cannot easily collect training data for every possible obstacle shape, distance, and mistake, so robots end up learning a few moves that only work in a narrow setup. This work starts from short clips of real human parkour moves, like stepping over, vaulting, climbing, and rolling. It uses motion matching, which is basically a smart “pick the next clip that fits best right now” search, to stitch those short clips into a long, smooth plan that looks like a human doing a whole course. Then it trains a controller with reinforcement learning (RL), which means the robot learns by trial and error to copy that plan while staying balanced and not falling. After training separate expert controllers for different moves, it compresses them into 1 controller that uses only onboard depth sensing and a simple “go this fast in this direction” command. In real tests on a Unitree G1 humanoid, it can clear multiple obstacles in a row, adapt when obstacles get moved, and climb a wall up to 1.25m.

This work makes a humanoid robot do simple parkour moves by looking with a depth camera and choosing the right move on the fly. The big deal is that it turns lots of small human moves into long, real-time robot behavior, without hand-coding every transition or retraining for each new course. A humanoid robot is usually good at steady walking, but it often fails when it has to do fast moves like jumping up, vaulting, or rolling, and then keep going to the next obstacle. The hard part is that you cannot easily collect training data for every possible obstacle shape, distance, and mistake, so robots end up learning a few moves that only work in a narrow setup. This work starts from short clips of real human parkour moves, like stepping over, vaulting, climbing, and rolling. It uses motion matching, which is basically a smart “pick the next clip that fits best right now” search, to stitch those short clips into a long, smooth plan that looks like a human doing a whole course. Then it trains a controller with reinforcement learning (RL), which means the robot learns by trial and error to copy that plan while staying balanced and not falling. After training separate expert controllers for different moves, it compresses them into 1 controller that uses only onboard depth sensing and a simple “go this fast in this direction” command. In real tests on a Unitree G1 humanoid, it can clear multiple obstacles in a row, adapt when obstacles get moved, and climb a wall up to 1.25m.

Rohan Paul

37,121 Aufrufe • vor 6 Monaten

Smart mapping for robots exploring huge areas! 🗺️ This is called Rolling-Bonxai. When a drone or ground robot maps a massive area, say a whole city or massive warehouse, storing the entire map in memory gets expensive fast. You run out of RAM. This solution is simple, only keep the nearby map in memory. As the robot moves, old chunks save to disk and new chunks load. Like how video games only render what's close to the player. The system divides the map into chunks (imagine a 3D grid of boxes). As the robot explores, it keeps chunks around it loaded in memory. Chunks it's moving away from get saved to disk. When it comes back, they load again. Check the GH repo here 👉🏼 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Smart mapping for robots exploring huge areas! 🗺️ This is called Rolling-Bonxai. When a drone or ground robot maps a massive area, say a whole city or massive warehouse, storing the entire map in memory gets expensive fast. You run out of RAM. This solution is simple, only keep the nearby map in memory. As the robot moves, old chunks save to disk and new chunks load. Like how video games only render what's close to the player. The system divides the map into chunks (imagine a 3D grid of boxes). As the robot explores, it keeps chunks around it loaded in memory. Chunks it's moving away from get saved to disk. When it comes back, they load again. Check the GH repo here 👉🏼 ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

15,646 Aufrufe • vor 1 Monat

Elon Musk: “Optimus Will Be The Biggest Product Ever Created” Let's see Optimus. You know, as you've heard me say before, I think Optimus will be the biggest product ever. But it is a very complex problem to solve. It's one of the hardest things to solve. To make a humanoid robot that can do tasks that you simply ask it to do, or show it a video, and it can do the task without any programming — like, no one's ever achieved this. And there are many challenges in the electromechanical design of the robot, to achieve sufficient dexterity and reliable hands, to be very reliable and have long, long wear and tear. But, I mean, it's good to be out in the field and not break down. Otherwise, you get this 70-kilogram robot that just flopped over, and you could carry it around like a baby, you know. And we don't want Optimus to go haywire. So, it's a lot of work to scale, to get the design right, and to scale production. And I really want to emphasize here that the production scaling challenge is very, very substantial. So this is going to be the hardest product to scale manufacturing that we've ever made at Tesla, because everything on the robot is new. And the difficulty of scaling the production up is proportionate to the newness of the parts in the robot.

Elon Musk: “Optimus Will Be The Biggest Product Ever Created” Let's see Optimus. You know, as you've heard me say before, I think Optimus will be the biggest product ever. But it is a very complex problem to solve. It's one of the hardest things to solve. To make a humanoid robot that can do tasks that you simply ask it to do, or show it a video, and it can do the task without any programming — like, no one's ever achieved this. And there are many challenges in the electromechanical design of the robot, to achieve sufficient dexterity and reliable hands, to be very reliable and have long, long wear and tear. But, I mean, it's good to be out in the field and not break down. Otherwise, you get this 70-kilogram robot that just flopped over, and you could carry it around like a baby, you know. And we don't want Optimus to go haywire. So, it's a lot of work to scale, to get the design right, and to scale production. And I really want to emphasize here that the production scaling challenge is very, very substantial. So this is going to be the hardest product to scale manufacturing that we've ever made at Tesla, because everything on the robot is new. And the difficulty of scaling the production up is proportionate to the newness of the parts in the robot.

Tesla Owners Silicon Valley

42,131 Aufrufe • vor 1 Monat

World Labs CEO Dr. Fei-Fei Li & SceniX Co-Founder Yunzhu Li on the data bottleneck in robotics and how world models help: "The lack of data in training, the lack of data in evaluation, this is very, very different from language models, where data is abundant on the internet." "And we know that in order for robotics to work, we have to somehow unlock the power of scaling law. But where does that come from?" "It's a profound problem that everybody's battling with in robotics." "We see a lot of unlock in being able to do this whole process through the modeling of the environments." "Being able to create these digital worlds... that's just going to unlock so much more potential for being able to replace all the costly and the unsafe data in the real environments with the data generated from the worlds for robots to be able to do scalable learning and evaluations." Fei-Fei Li Yunzhu Li martin_casado

World Labs CEO Dr. Fei-Fei Li & SceniX Co-Founder Yunzhu Li on the data bottleneck in robotics and how world models help: "The lack of data in training, the lack of data in evaluation, this is very, very different from language models, where data is abundant on the internet." "And we know that in order for robotics to work, we have to somehow unlock the power of scaling law. But where does that come from?" "It's a profound problem that everybody's battling with in robotics." "We see a lot of unlock in being able to do this whole process through the modeling of the environments." "Being able to create these digital worlds... that's just going to unlock so much more potential for being able to replace all the costly and the unsafe data in the real environments with the data generated from the worlds for robots to be able to do scalable learning and evaluations." Fei-Fei Li Yunzhu Li martin_casado

a16z

75,232 Aufrufe • vor 1 Monat

A GUY PUT GPT-LEVEL AI INSIDE A 3D PRINTED ROBOT AND RELEASED THE ENTIRE THING FOR FREE Most people look at a 3D printer and see plastic toys This guy printed a robot with expressive eyes, object tracking, image recognition, support for ChatGPT, Qwen and offline models, then uploaded the STL files, code and hardware designs for anyone to build themselves The interesting part isn't the robot It's how fast the gap between "I have an idea" and "I built a working product" is disappearing A few years ago this required a team of engineers Now one person with a printer, open-source software and enough curiosity can build something that looks like it came from a robotics startup We're getting very close to a world where manufacturing, AI and robotics stop being separate industries and start becoming the same thing

A GUY PUT GPT-LEVEL AI INSIDE A 3D PRINTED ROBOT AND RELEASED THE ENTIRE THING FOR FREE Most people look at a 3D printer and see plastic toys This guy printed a robot with expressive eyes, object tracking, image recognition, support for ChatGPT, Qwen and offline models, then uploaded the STL files, code and hardware designs for anyone to build themselves The interesting part isn't the robot It's how fast the gap between "I have an idea" and "I built a working product" is disappearing A few years ago this required a team of engineers Now one person with a printer, open-source software and enough curiosity can build something that looks like it came from a robotics startup We're getting very close to a world where manufacturing, AI and robotics stop being separate industries and start becoming the same thing

lagerskoy

57,099 Aufrufe • vor 2 Monaten

Another piece from the Sasha Rush conversation, this time on rewards for coding RL. He said Cursor uses a mix. Some rewards look at the tool calls themselves, some only at the final output. Everything end-to-end, no process rewards guessing what happens in the middle. I agree with him that the next push (and the current one) is going past verifiers. Coding is a good example where you get both. You have the 0/1 signals, but also a ton of soft ones. Does the diff read well? Did the agent burn context for no reason? Is the PR the right size? Did it rewrite a file it didn't need to touch? So many signals sitting there, and we need to figure out how to actually train on them. This is also why I think the environment (and the data) is becoming the bottleneck, not the algorithm. GRPO works. PPO works. The hard part is building a sandbox that exposes enough signal to learn from. We're already seeing this, and it'll only accelerate: a whole category of companies selling environments is emerging, and they have a good reward to do so…

Another piece from the Sasha Rush conversation, this time on rewards for coding RL. He said Cursor uses a mix. Some rewards look at the tool calls themselves, some only at the final output. Everything end-to-end, no process rewards guessing what happens in the middle. I agree with him that the next push (and the current one) is going past verifiers. Coding is a good example where you get both. You have the 0/1 signals, but also a ton of soft ones. Does the diff read well? Did the agent burn context for no reason? Is the PR the right size? Did it rewrite a file it didn't need to touch? So many signals sitting there, and we need to figure out how to actually train on them. This is also why I think the environment (and the data) is becoming the bottleneck, not the algorithm. GRPO works. PPO works. The hard part is building a sandbox that exposes enough signal to learn from. We're already seeing this, and it'll only accelerate: a whole category of companies selling environments is emerging, and they have a good reward to do so…

Ravid Shwartz Ziv

14,826 Aufrufe • vor 4 Monaten