Pinned Tweet
This is the most impressive result I’ve seen in my PhD.
One policy, trained from scratch, solves all 20 NIST assembly tasks. It can even complete 10 tasks in a row, with a single binary reward only after all 10 are done.
@octi_zhang combined diverse resets from OmniReset, task multi-task training to unlock finer dexterity and longer horizons. No extra GPUs needed: one run on the same hardware replaces 20 specialist runs.
Videos and details: alpha-nist.github.io
Proud to be part of this incredible work! More in his thread 👇
One policy, trained from scratch, solves all 20 NIST assembly tasks. It can even complete 10 tasks in a row, with a single binary reward only after all 10 are done.
@octi_zhang combined diverse resets from OmniReset, task multi-task training to unlock finer dexterity and longer horizons. No extra GPUs needed: one run on the same hardware replaces 20 specialist runs.
Videos and details: alpha-nist.github.io
Proud to be part of this incredible work! More in his thread 👇
(1/6) From @GoogleDeepmind’s AlphaGo to AlphaZero, we saw machines find moves that reached beyond what we’d imagined for them. No one taught them those strategies. They emerged from reinforcement learning: playing, failing, and trying again under a single binary task reward for pired by them, we couldn’t help but wonder: what might a robot discover if we gave it that same freedom to learn?
That question led us to AlphaNIST. Starting tabula rasa, with that same binary reward plus only a simple control cost, a single policy solves all 20 @NIST assembly tasks at a 98.2% success rate in simulation. More surprising still, AlphaNIST can complete 10 different tasks in a row, receiving that reward only once, after all 10 are done.
Just as AlphaGo’s strategies surprised its creators, AlphaNIST discovered dexterous strategies we couldn’t have come up with ourselves.
Paper: arxiv.org/abs/2610.13354
Website: alpha-nist.github.io
18
59
883
43,217
Patrick Yin ✈️ CoRL retweeted
Excited to share the first paper of my undergrad, Q-OIL, which we’ll be presenting at #CoRL2026 in Austin!
Interventions provide a natural source of feedback for robots to learn from in the real world. But, humans intervene suboptimally, which needs to be accounted for in both and algorithms.
We analyze human intervention data and measure intervention suboptimality in both timing and execution. Thus, rather than using interventions as supervision, we use them as soft exploration guidance.
(1/8)
Interventions provide a natural source of feedback for robots to learn from in the real world. But, humans intervene suboptimally, which needs to be accounted for in both and algorithms.
We analyze human intervention data and measure intervention suboptimality in both timing and execution. Thus, rather than using interventions as supervision, we use them as soft exploration guidance.
(1/8)
13
35
152
16,841
Great work with @memmelma and team! D&A takes robot videos and gives densified reward for downstream learning. Here's them doing it on our SGS nut twirling task from sgs-rl.github.io
Want to turn any VLM into a better reward model for robotics? (yes, even Astra)
Introducing D&A: decoupling description and assessment improves any VLM family for zero-shot robot rewards.
D&A scales with improvements across GPT, Gemini, Claude, Qwen, ...
It enables zero-shot RL and fde-as-policies via rich language feedback, and achieves SOTA on three benchmarks.
Website: weirdlabuw.github.io/dna
3
7
39
2,312
Patrick Yin ✈️ CoRL retweeted
Our new robot OpenDubs worked a 6 hour shift autonomously handing out candy to trick-or-treaters for Halloween this year @uwcse 🤖🍬🎃
Thanks to the many folks who made this fun side project into a reality 😄 We took quite the full stack approach to this endeavor - literally m the ground up in just under a month and a half! @sidharthtalia @arc_hitter @siddhss5 @uw_robotics
Stay tuned to see more of what OpenDubs can do! 🦾
Thanks to the many folks who made this fun side project into a reality 😄 We took quite the full stack approach to this endeavor - literally m the ground up in just under a month and a half! @sidharthtalia @arc_hitter @siddhss5 @uw_robotics
Stay tuned to see more of what OpenDubs can do! 🦾
2
9
48
3,274
really glad to see reset density studied more deeply. and more robot embodiments! microduck ftw nice work @corbraun and Tilman
Finding feasible robot states for goal-conditioned RL is one thing. Sampling as diverse states as possible is another.
Diverse resets and goals help GCRL, but contact and collision constraints make truly uniform sampling hard.
Our NeurIPS ’26 work introduces MASEM. 🧵👇.
0
0
9
992
Patrick Yin ✈️ CoRL retweeted
Every robot already carries contact sensors: its own joints. When a robot touches something, its joints drift slightly off their commands, and the encoders pick it up. Legged robots and humanoids already leverage this signal to walk blind - but is it enough for blind precise it is! Our policies can insert pegs, mesh gears, and thread nuts using only the robot's joint positions - no camera, no tactile, and no object poses. And the peg policy works zero-shot in the real world! 🧵
w/ my amazing co-first author @younghyo_park, and advised by @kushalk_ and @pulkitology.
w/ my amazing co-first author @younghyo_park, and advised by @kushalk_ and @pulkitology.
18
54
425
36,350
Patrick Yin ✈️ CoRL retweeted
Can a robot assemble tight-clearance parts using only its proprioception? No camera, no tactile sensor, no F/T sensor, no joint torque sensing, no priors over the object geometries or poses: just joint position readings.
It actually can! Blind policies trained in simulation , and thread nuts, and the peg policy works zero-shot in the real world. Here's what we found 🧵
Joint work with @BronarsToni, my wonderful co-first author, who's on the job market right now! Advised by @kushalk_ and @pulkitology.
It actually can! Blind policies trained in simulation , and thread nuts, and the peg policy works zero-shot in the real world. Here's what we found 🧵
Joint work with @BronarsToni, my wonderful co-first author, who's on the job market right now! Advised by @kushalk_ and @pulkitology.
24
62
535
45,942
Patrick Yin ✈️ CoRL retweeted
Exploration is hard for RL when training high-DoF dex hands ✋
X-Reset makes generalist object-centric RL easier by resetting to human hand-object states 🙋♀️
We mitigate the exploration problem, and train a single policy to manipulate a variety of objects to any goal pose 🧵
X-Reset makes generalist object-centric RL easier by resetting to human hand-object states 🙋♀️
We mitigate the exploration problem, and train a single policy to manipulate a variety of objects to any goal pose 🧵
3
18
127
15,761
Patrick Yin ✈️ CoRL retweeted
Three AlphaNIST findings I'm particularly excited about:
1. Task diversity can make difficult tasks easier to learn.
2. Dexterity isn't just about more fingers.
3. Long-horizon coordination emerges without explicit planning.
All with binary task reward!
See @octi_zhang 's thread
1. Task diversity can make difficult tasks easier to learn.
2. Dexterity isn't just about more fingers.
3. Long-horizon coordination emerges without explicit planning.
All with binary task reward!
See @octi_zhang 's thread
(1/6) From @GoogleDeepmind’s AlphaGo to AlphaZero, we saw machines find moves that reached beyond what we’d imagined for them. No one taught them those strategies. They emerged from reinforcement learning: playing, failing, and trying again under a single binary task reward for pired by them, we couldn’t help but wonder: what might a robot discover if we gave it that same freedom to learn?
That question led us to AlphaNIST. Starting tabula rasa, with that same binary reward plus only a simple control cost, a single policy solves all 20 @NIST assembly tasks at a 98.2% success rate in simulation. More surprising still, AlphaNIST can complete 10 different tasks in a row, receiving that reward only once, after all 10 are done.
Just as AlphaGo’s strategies surprised its creators, AlphaNIST discovered dexterous strategies we couldn’t have come up with ourselves.
Paper: arxiv.org/abs/2610.13354
Website: alpha-nist.github.io
0
5
17
996
Patrick Yin ✈️ CoRL retweeted
Amazing to see this level of dexterity emerging from a binary success / fail reward. Great work @octi_zhang !
(1/6) From @GoogleDeepmind’s AlphaGo to AlphaZero, we saw machines find moves that reached beyond what we’d imagined for them. No one taught them those strategies. They emerged from reinforcement learning: playing, failing, and trying again under a single binary task reward for pired by them, we couldn’t help but wonder: what might a robot discover if we gave it that same freedom to learn?
That question led us to AlphaNIST. Starting tabula rasa, with that same binary reward plus only a simple control cost, a single policy solves all 20 @NIST assembly tasks at a 98.2% success rate in simulation. More surprising still, AlphaNIST can complete 10 different tasks in a row, receiving that reward only once, after all 10 are done.
Just as AlphaGo’s strategies surprised its creators, AlphaNIST discovered dexterous strategies we couldn’t have come up with ourselves.
Paper: arxiv.org/abs/2610.13354
Website: alpha-nist.github.io
2
3
26
1,862
Patrick Yin ✈️ CoRL retweeted
this is super cool, multi-task RL for dexterous assembly!
I'm surprised that multi-task training shows positive transfer between tasks with only 20 tasks, perhaps it's because the skills required are similar? Either way, makes the case for sim rl stronger 😁
I'm surprised that multi-task training shows positive transfer between tasks with only 20 tasks, perhaps it's because the skills required are similar? Either way, makes the case for sim rl stronger 😁
(4/6) We started this project with one goal: solve all 20 NIST assembly tasks with RL. Our earlier work on Success-Guided Sampling (SGS) had already solved one of them, so we reused its recipe: a binary task reward with a simple control cost, a broad reset distribution from task sampler. Training a specialist policy for each task got us 17 of the 20. The last three, the 4 mm peg, 4 mm rod, and M4 nut, wouldn’t budge. Their policies couldn’t even pick the parts off the table: a slight miss with the gripper pushed these tiny parts away instead of pinching them.
The obvious fix was to design easier resets or denser rewards for those three until they learned. Instead, we wondered: could skills learned on the other 17 tasks help with the hard grasps the last three needed? So we tried what seemed like the harder path: one policy trained on all 20 tasks at once, with each episode starting from any of the 20 parts.
Then the magic happened. The generalist learned to pick up and assemble all three tiny parts, alongside the other 17. Multi-task training, the seemingly harder path, made the hardest tasks easier to learn.
0
2
20
983
Patrick Yin ✈️ CoRL retweeted
@octi_zhang and team with some pretty cool new results :)
(1/6) From @GoogleDeepmind’s AlphaGo to AlphaZero, we saw machines find moves that reached beyond what we’d imagined for them. No one taught them those strategies. They emerged from reinforcement learning: playing, failing, and trying again under a single binary task reward for pired by them, we couldn’t help but wonder: what might a robot discover if we gave it that same freedom to learn?
That question led us to AlphaNIST. Starting tabula rasa, with that same binary reward plus only a simple control cost, a single policy solves all 20 @NIST assembly tasks at a 98.2% success rate in simulation. More surprising still, AlphaNIST can complete 10 different tasks in a row, receiving that reward only once, after all 10 are done.
Just as AlphaGo’s strategies surprised its creators, AlphaNIST discovered dexterous strategies we couldn’t have come up with ourselves.
Paper: arxiv.org/abs/2610.13354
Website: alpha-nist.github.io
0
6
58
3,258
Patrick Yin ✈️ CoRL retweeted
Simulation will be a bitter lesson for human data folks 😁
(1/6) From @GoogleDeepmind’s AlphaGo to AlphaZero, we saw machines find moves that reached beyond what we’d imagined for them. No one taught them those strategies. They emerged from reinforcement learning: playing, failing, and trying again under a single binary task reward for pired by them, we couldn’t help but wonder: what might a robot discover if we gave it that same freedom to learn?
That question led us to AlphaNIST. Starting tabula rasa, with that same binary reward plus only a simple control cost, a single policy solves all 20 @NIST assembly tasks at a 98.2% success rate in simulation. More surprising still, AlphaNIST can complete 10 different tasks in a row, receiving that reward only once, after all 10 are done.
Just as AlphaGo’s strategies surprised its creators, AlphaNIST discovered dexterous strategies we couldn’t have come up with ourselves.
Paper: arxiv.org/abs/2610.13354
Website: alpha-nist.github.io
9
20
496
38,167
Patrick Yin ✈️ CoRL retweeted
@octi_zhang is a legend. the multitask policy learning super impressive
(1/6) From @GoogleDeepmind’s AlphaGo to AlphaZero, we saw machines find moves that reached beyond what we’d imagined for them. No one taught them those strategies. They emerged from reinforcement learning: playing, failing, and trying again under a single binary task reward for pired by them, we couldn’t help but wonder: what might a robot discover if we gave it that same freedom to learn?
That question led us to AlphaNIST. Starting tabula rasa, with that same binary reward plus only a simple control cost, a single policy solves all 20 @NIST assembly tasks at a 98.2% success rate in simulation. More surprising still, AlphaNIST can complete 10 different tasks in a row, receiving that reward only once, after all 10 are done.
Just as AlphaGo’s strategies surprised its creators, AlphaNIST discovered dexterous strategies we couldn’t have come up with ourselves.
Paper: arxiv.org/abs/2610.13354
Website: alpha-nist.github.io
0
3
14
430
Patrick Yin ✈️ CoRL retweeted
At last… torturous, terrifyingly tiny tasks have been solved by RL.
The M4 nut, 4mm peg, and 4mm rod were too small for any single-task policy to learn. Training on all 20 NIST tasks at once enabled success on all of them. And the same policy can do 10 tasks back-to-back, with very end.
Videos and details: alpha-nist.github.io
More in @octi_zhang’s thread
The M4 nut, 4mm peg, and 4mm rod were too small for any single-task policy to learn. Training on all 20 NIST tasks at once enabled success on all of them. And the same policy can do 10 tasks back-to-back, with very end.
Videos and details: alpha-nist.github.io
More in @octi_zhang’s thread
(3/6) We also never told it what order to work in. Over long horizons, AlphaNIST began organizing its own work: bringing parts within reach, moving obstructions aside, and returning to pieces it had left unfinished. It received a single binary reward only when the entire complete, so the task order and strategy emerged purely from RL.
0
3
14
1,123
Patrick Yin ✈️ CoRL retweeted
What does RL learn from 10 assembly tasks in a row, with sparse reward only at the very end of all 10 tasks? Traditionally, absolutely nothing!
That’s 1,125 actions over 45 seconds. Even if a random policy somehow nails each task 1% of the time, chaining all 10 happens with e odds than winning Powerball twice in a row.
Excited to share AlphaNIST, a single policy which achieves 98.2% success rate across all tasks, and is able to solve 10 assemblies in a row under sparse reward! Check out our website: alpha-nist.github.io.
More details in @octi_zhang ‘s thread, and some extra comments of my own below! 🧵
That’s 1,125 actions over 45 seconds. Even if a random policy somehow nails each task 1% of the time, chaining all 10 happens with e odds than winning Powerball twice in a row.
Excited to share AlphaNIST, a single policy which achieves 98.2% success rate across all tasks, and is able to solve 10 assemblies in a row under sparse reward! Check out our website: alpha-nist.github.io.
More details in @octi_zhang ‘s thread, and some extra comments of my own below! 🧵
(1/6) From @GoogleDeepmind’s AlphaGo to AlphaZero, we saw machines find moves that reached beyond what we’d imagined for them. No one taught them those strategies. They emerged from reinforcement learning: playing, failing, and trying again under a single binary task reward for pired by them, we couldn’t help but wonder: what might a robot discover if we gave it that same freedom to learn?
That question led us to AlphaNIST. Starting tabula rasa, with that same binary reward plus only a simple control cost, a single policy solves all 20 @NIST assembly tasks at a 98.2% success rate in simulation. More surprising still, AlphaNIST can complete 10 different tasks in a row, receiving that reward only once, after all 10 are done.
Just as AlphaGo’s strategies surprised its creators, AlphaNIST discovered dexterous strategies we couldn’t have come up with ourselves.
Paper: arxiv.org/abs/2610.13354
Website: alpha-nist.github.io
7
14
158
8,793
Beyond the collaborators @octi_zhang thanks in his thread, special thanks to Karl Van Wyk, who invented the @NIST board, and to Joe Falco and Kenneth Kimble for their work on it since. None of this would exist without them.
0
0
4
753
Patrick Yin ✈️ CoRL retweeted
Failures are inevitable for robots that are deployed in the *big* world. We introduce ReSYNC, a system that leverages large-scale RL sim2real to recover from local failures and concept discovery to explain/avoid new failures. Check it out here: jaraxxus-me.github.io/ReSYNC/
13
28
87
8,253
Patrick Yin ✈️ CoRL retweeted
since people seemed to enjoy our locomotion results, chef 🧑🍳@octi_zhang cooked up a special treat à la SGS on a different embodiment (a wheeled @UnitreeRobotics robot) to celebrate the release! The robot is placed in this big world, and its job is to reach the yellow dots.
(1/5) Everyone says dexterous manipulation needs real-world data. These policies were trained with none. Our robots mesh gears, thread nuts, and insert pegs zero-shot from RGB, and the same recipe works for locomotion across terrains much harder than before.
The missing piece rithm, but a small tweak to what the robot practices.
Today we're releasing Success-Guided Sampling (SGS), a simple change to the task sampler that unlocks new levels of dexterity and agility with large-scale, end-to-end sim2real RL.
No demonstrations, no per-task reward tuning. Just PPO at scale.
Check out the policies yourself at sgs-rl.github.io! All videos at 1x speed! 🧵 🔈
4
14
105
4,580
Patrick Yin ✈️ CoRL retweeted
The value of scaled RL training in simulation can't be understated. Huge congratulations to @octi_zhang and everyone else who worked on the SGS paper. What comes next is even cooler - stay tuned!
(1/5) Everyone says dexterous manipulation needs real-world data. These policies were trained with none. Our robots mesh gears, thread nuts, and insert pegs zero-shot from RGB, and the same recipe works for locomotion across terrains much harder than before.
The missing piece rithm, but a small tweak to what the robot practices.
Today we're releasing Success-Guided Sampling (SGS), a simple change to the task sampler that unlocks new levels of dexterity and agility with large-scale, end-to-end sim2real RL.
No demonstrations, no per-task reward tuning. Just PPO at scale.
Check out the policies yourself at sgs-rl.github.io! All videos at 1x speed! 🧵 🔈
3
10
53
8,651