←
Joseph Amigo41 posts
Joseph Amigo avatar

Joseph Amigo

@Jsphamigo

PhD Candidate at NYU & LAAS-CNRS - Robotics, Reinforcement Learning, Deep Learning

◷ Joined February 2018▧ 3 Photos and videos
Kyle🤖🚀🦭 avatar
⟳ Joseph Amigo retweeted
Kyle🤖🚀🦭@KyleMorgenstein·21h

people kept asking me for “hot takes” at CoRL. I was sorry to disappoint, because what I came up with felt lukewarm: RL is the only thing that has ever worked at scale. exploration + adaptation. that’s it. everything else is a party trick. RL is the ultimate bitter lesson.

you may think “RL is not the right tool for this problem” and you would be wrong. it’s all RL. this is sick.

1:10
Joseph Amigo avatar
Joseph Amigo@Jsphamigo·Nov 16

Awesome work on MABEL, @thejerrycheng, super cool to see everything open-sourced.

Today, I'm excited to finally introduce MABEL — our fully open-source, mobile bimanual robot! 🤖🦾 $8K in parts. Just 2 weeks from parts to a fully tested robot. And we're open-sourcing EVERYTHING: hardware, firmware, teleop pipeline, controllers, learning pipelines, and digital twin simulation. Built by us at @nyuniversity and @UofTRobotics in 3 months, started in a tiny NYC studio apartment 🗽🍕 Our goal is simple: make capable, full-stack robotics accessible to everyone. 🚀 🧵👇 (1/7)

Post attachmentPost attachment
0:05
0:41
Joseph Amigo avatar
Joseph Amigo@Jsphamigo·Nov 14

We never properly announced it: our paper was accepted at #CoRL2026 🎉 We train robot policies from scratch entirely inside a large world model, with first-order gradients through a lightweight latent surrogate. coupled-global-local-wm-rl.p…

Joseph Amigo avatar
Joseph Amigo@Jsphamigo·Nov 14

The FAA extended my #CoRL2026 by three days 🥲 I'm in Austin until Monday. If you're still around and want to talk world models, RL or our CoRL paper, let's grab coffee or BBQ 🤠 This G1 policy was trained entirely inside a learned world model, then deployed zero-shot 👇

0:50
Chris Paxton avatar
⟳ Joseph Amigo retweeted
Chris Paxton@chris_j_paxton·Nov 12

ChatGPT required RL, it was not pure imitation learning or dagger; reasoning models also required RL. Obviously the same will be true of robotics, the question is just if base models are good enough

pendulum has fully swung back from where it was last year (“scale up data collection, behavior clone large models”). everyone is finally seeing the shining light of RL!

Chris Paxton avatar
⟳ Joseph Amigo retweeted
Chris Paxton@chris_j_paxton·Nov 6

CoRL has become the place to be if you are interested in robotics. There's no robotics without intelligence -- robots will never scale without scalable learning for problems like contact-rich mobile manipulation -- and so we've seen this academic conference become so essential IMO. Looking forward to meeting people there

CoRL is now sold out. We have an incredible program lined up. Can't wait to see y'all in Austin next week! corl.org/program

Post attachment
Jerry Chéng avatar
⟳ Joseph Amigo retweeted
Jerry Chéng@thejerrycheng·Nov 4

Today, I'm excited to finally introduce MABEL — our fully open-source, mobile bimanual robot! 🤖🦾 $8K in parts. Just 2 weeks from parts to a fully tested robot. And we're open-sourcing EVERYTHING: hardware, firmware, teleop pipeline, controllers, learning pipelines, and digital twin simulation. Built by us at @nyuniversity and @UofTRobotics in 3 months, started in a tiny NYC studio apartment 🗽🍕 Our goal is simple: make capable, full-stack robotics accessible to everyone. 🚀 🧵👇 (1/7)

Post attachmentPost attachment
0:05
0:41
Qiayuan Liao avatar
⟳ Joseph Amigo retweeted
Qiayuan Liao@qiayuanliao·Oct 28

I think your claim conflicts with both SuperMimic and DMO, @Jsphamigo. In SuperMimic, we have an ablation showing that APG is better with the common standard whole-body tracking setup. DMO has a more interesting setup. It has the same env/task as RWM (quadrupedal locomotion without reference motions). But for sampling and gradients, it does the exact opposite of RWM: the forward rollout uses simulation, but the backward pass uses the first-order gradients from a learned network. Indeed, we might need to run more controlled experiments.

Qiayuan Liao avatar
⟳ Joseph Amigo retweeted
Qiayuan Liao@qiayuanliao·Oct 26

Just realized that we didn’t discuss the DMO in our paper! The DMO provided lots of insight when we were working at SuperMimic. It tells us how powerful a first-order gradient is by construction, which can complete the only missing term in the 2x2 ablation grid. That is {PPO, APG}x{sim, WM}.

Great to see first-order gradients through learned world models scale to whole-body tracking. Congrats @qiayuanliao et al.! 👏 We explored a related, decoupled variant in DMO (CoRL '25): forward in the real simulator, backward through a learned model. machines-in-motion.github.io… ⁠

0:16
Joseph Amigo avatar
Joseph Amigo@Jsphamigo·Oct 26

For context: DMO (CoRL '25) showed the same two effects on a single V100. First-order gradients through a learned model needed far fewer environment steps than zeroth-order RL (PPO/SAC) and converged faster in wall-clock time 👇

Post attachmentPost attachment
Joseph Amigo avatar
Joseph Amigo@Jsphamigo·Oct 26

Great to see first-order gradients through learned world models scale to whole-body tracking. Congrats @qiayuanliao et al.! 👏 We explored a related, decoupled variant in DMO (CoRL '25): forward in the real simulator, backward through a learned model. machines-in-motion.github.io… ⁠

0:16

We found a way to scale humanoid control that reduces GPU-hours by more than 1000×. Across 129,514 motions, SuperMimic surpasses the state of the art (128 GPUs, 7 days) with 1 GPU in 3 hours. Project: supermimic.github.io Demo: supermimic.github.io/demo/

1:14
Chenhao Li 📌 CoRL avatar
⟳ Joseph Amigo retweeted
Chenhao Li 📌 CoRL@breadli428·Oct 25

It’s great to see growing interest in MBRL. Last year in RWM, we demonstrated that policies trained entirely in world-model imagination transfer zero-shot to the real world. Sometimes, ideas are ahead of their time. Glad to see this direction getting the recognition it deserves.

🧠Model-Based RL shows promises but has seen limited success in real-world robotics. 🌎Introducing Robotic World Model, a black-box end-to-end neural dynamics model that bridges this gap, where policies are trained purely in imagination. @NeurIPSConf 🎯sites.google.com/view/roboti…

2:40
Dyuman Aditya avatar
⟳ Joseph Amigo retweeted
Dyuman Aditya@dyumanaditya·Sep 28

🤖 Ever wondered if you can train end-to-end humanoid policies entirely within a differentiable simulator? 🥁 Introducing 𝗕𝘂𝗻𝗱𝗹𝗲𝗱 𝗖𝗼𝗻𝘁𝗮𝗰𝘁 𝗚𝗿𝗮𝗱𝗶𝗲𝗻𝘁𝘀 (𝗕𝗖𝗚) 🌐: bundledcontactgradients.gith… 📄: arxiv.org/abs/2609.30951 🧵👇

0:20
Unitree avatar
⟳ Joseph Amigo retweeted
Unitree@UnitreeRobotics·Sep 7

Unitree Breakthrough: The World’s First Real-Time World Model-Driven Fully Autonomous Humanoid Robot Combat🥊 UnifoLM-X2-1.0 breaks through world-action foundation models' bottlenecks in instant planning, decision-making, and dynamic interactive execution, achieve high dynamics, strong interaction, real-time prediction and planning of the future, achieve fully autonomous combat for humanoid robots. This validates the fundamental feasibility of large-scale deployment of world model-driven humanoid robots.

0:38
Jerry Chéng avatar
⟳ Joseph Amigo retweeted
Jerry Chéng@thejerrycheng·Sep 4

Come checkout my lab mates great work on using world model for RL policy! They trained their interactive video model completely from scratch - the best I have ever tried. coupled-global-local-wm-rl.p…

Reinforcement learning has unlocked tremendous gains in broader AI and humanoid robot whole body control, but still has yet to deliver strong results for robot manipulation due to the difficulty of capturing tasks in an accurate simulation; one way to fix this is to create complex environments with learned world models. But learning fully within world models is often intractable as powerful world models are computationally too expensive. Instead, @Jsphamigo and @Rk4342R propose to break the problem down: into a large-scale world model that can generate forward trajectories, and a lightweight, low-dimensional latent-space model which can approximate local dynamics of a problem, without needing to back-propagate through a heavy global model. What this means: you can do reinforcement learning in this coupled local-global world model, and learn contact rich skills for a robot. To learn more, watch Episode #101 of RoboPapers, with @micoolcho and @DJiafei!

51:26
Frontier Signal avatar
⟳ Joseph Amigo retweeted
Frontier Signal@FrontierSignal_·Sep 4

Clever split, but the local model still needs feeding. Contact dynamics come from real contact hours, so the bottleneck moves from simulator fidelity to collecting messy physical data.