@mo_danesh

phd @McGillU and @mila_quebec, working on rl and robot learning

 Montréal, Québec
modanesh.github.io
Joined February 2017
  • Tweets637
  • Following847
  • Followers294
78Media
📢 New paper out! We introduce QWM: a single locomotion world model trained across 8 quadrupeds and deployed zero-shot on robots it had never seen by conditioning on their morphology specs: ANYmal-D and Unitree Go1 🦾 No fine-tuning, no warm-up, no retraining from scratch. The key insight: robot morphology isn't a latent variable to infer from motion history, it's a known engineering spec sitting in the USD (or URDF) file. So we just use it directly.
1
3
1
24
3,085
my co-first author at NeurIPS didn't get the invitation from the waitlist to register meaning he can't attend a conference he has a paper in. wondering who got it. i thought the whole point of having 3 locations was to accommodate everyone. apparently publishing a paper doesn't guarantee you can actually attend your own conference. really frustrating and ridiculous!
17
2,464
🎉 The CoRL 2026 accepted papers are out! 📄 687 papers 🎤 21 orals ✨ 72 spotlights See you in Austin, Nov 9–11. Congrats to all authors 👏 corl.org/program/accepted-pa… #CoRL2026

CoRL 2026 - Accepted Papers

corl.org
1
6
1
83
7,256
100% modanesh.github.io/papers/qw…
Chenhao Li@breadli428Oct 15
Make RL inside world models great again! sites.google.com/view/uncert…

Uncertainty-Aware RWM

sites.google.com
3
133
Hence the question: which is better: a roboticist who learns AI, or an AI person who learns robotics?
C. Zhang@ChongZzZhangOct 13
No more robotics, it's all AI
1
2
710
Mohamad H. Danesh retweeted
Robotic Systems Lab@leggedroboticsOct 10
In our latest work at @corl_conf, we introduce Quadrupedal World Model: a single morphology-conditioned dynamics model, where policies are trained entirely in imagination. We found generalization emerges by conditioning the dynamics, not the policy. 🔗modanesh.github.io/papers/qw…
Mohamad H. Danesh
@mo_daneshOct 7
What if one world model could control many different robots? 🤖🌍 That's QWM. In this 2 minutes video we show keypoints about QWM: learning across heterogeneous quadrupeds, and transferring zero-shot to unseen morphologies. No fine-tuning. No warm-up. No retraining. Let's talk it further at CoRL 🥳
3
7
2
57
4,790
Mohamad H. Danesh retweeted
Chenhao Li@breadli428Oct 10
🌐Does modeling dynamics provide a better path to generalization than learning model-free policies alone? 🐕In QWM @corl_conf, a morphology-conditioned world model learns shared locomotion dynamics and enables zero-shot transfer to unseen morphologies. 🔗modanesh.github.io/papers/qw…
Mohamad H. Danesh
@mo_daneshOct 7
What if one world model could control many different robots? 🤖🌍 That's QWM. In this 2 minutes video we show keypoints about QWM: learning across heterogeneous quadrupeds, and transferring zero-shot to unseen morphologies. No fine-tuning. No warm-up. No retraining. Let's talk it further at CoRL 🥳
2
8
62
3,493
What if one world model could control many different robots? 🤖🌍 That's QWM. In this 2 minutes video we show keypoints about QWM: learning across heterogeneous quadrupeds, and transferring zero-shot to unseen morphologies. No fine-tuning. No warm-up. No retraining. Let's talk it further at CoRL 🥳
1
5
3
19
8,245
Checkout the website here: 🔗 modanesh.github.io/papers/qw…

Morphology-Conditioned World Model for Cross-Embodiment Quadrupedal Locomotion

A morphology-conditioned world model that trains quadruped policies in imagination and transfers them zero-shot to unseen robots, in sim and on hardware.

modanesh.github.io
2
165
Singularity is here
OpenAI
@OpenAIOct 6
We’re releasing a broad range of new mathematical results produced by an internal frontier model. We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results. github.com/openai/math

GitHub - openai/math

Contribute to openai/math development by creating an account on GitHub.

github.com
3
149
🚀 Hetero-IsaacLab v0.3 is out! Expanded the framework beyond quadrupeds to support heterogeneous humanoids as well: Unitree H1, Unitree H2, Unitree G1, Cassie, Digit You can now train policies across different morphologies in a signle heterogenous environment: 🌐github.com/modanesh/Hetero-I…
4
125
Mohamad H. Danesh retweeted
Anas Houssaini
@aahoussainiSep 24
Drift Q-Learning just got into #Neurips26 ! 🇦🇺 I guess see you in Australia!! Project Page: driftql.github.io/
1
1
7
241
DriftQL 🤝 NeurIPS! Sydney, here we come! 🇦🇺
1
25
2,045
Mohamad H. Danesh retweeted
Agility
@agilityroboticsSep 17
For years, Digit had a bird-like stance, tuned for efficient, dynamic motion. Digit 5 is being engineered with a redesigned leg for the other half of real work: squatting low and lifting heavy, the way a person does when the load's on the floor. Thanks to @bheater and @a3automate for the coverage. automate.org/robotics/indust…
13
22
2
148
8,553
No more bird legs
Humanoids daily
@humanoidsdailySep 15
🚨BREAKING: Agility unveils Digit 5. The redesigned humanoid has a more humanlike leg profile and is built for what Agility calls “cooperatively safe” work—operating near people without physical safety barriers.
122
Mohamad H. Danesh retweeted
Glen Berseth@GlenBersethSep 8
At this year's Montreal Robotics Summer School, teams programmed a Unitree Go1 to walk itself to dinner — sim2real RL, vision-based nav, AprilTags and all. Congrats to our Winning teams (5 & 6)! Full recap 👉 fracturedplane.com/blog/2026… #Robotics #ReinforcementLearning #Mila
2
15
1,823
Accepted to CoRL! 🥳 See you in Austin!
Mohamad H. Danesh
@mo_daneshMay 1
📢 New paper out! We introduce QWM: a single locomotion world model trained across 8 quadrupeds and deployed zero-shot on robots it had never seen by conditioning on their morphology specs: ANYmal-D and Unitree Go1 🦾 No fine-tuning, no warm-up, no retraining from scratch. The key insight: robot morphology isn't a latent variable to infer from motion history, it's a known engineering spec sitting in the USD (or URDF) file. So we just use it directly.
2
1
15
1,220
Mohamad H. Danesh retweeted
Chenhao Li@breadli428Aug 24
Mind blowing 🤯 This is killing research in many academic labs. Nor do I believe any published research was ever close to this, regardless of the posted videos. Does it still make sense to work on sim-to-real locomotion?
Takumi Kawasetsu, 川節拓実
@takumi_k_jpnAug 24
100m障害物走の決勝
18
26
1
341
82,072
Mohamad H. Danesh retweeted
Chenhao Li@breadli428Aug 15
Great talk by Scott Fujimoto at our World Models + Model-Based RL workshop explaining what world models are. A great day of talks going on @UMontreal 🇨🇦! #RLC2026
Chenhao Li@breadli428Aug 14
I’m at Reinforcement Learning Conference in Montreal this week! 🇨🇦 Come to our Model-Based RL workshop tomorrow! Our speakers are cool!!
2
7
106
9,244
Our workshop on world modeling in RL just started! Come join us to hear the latest on world modeling from the experts and have productive discussions in our poster sessions. 📍 B-4335, 3200 Jean-Brillant, Université de Montréal
6
161
Mohamad H. Danesh retweeted
Chenhao Li@breadli428Aug 14
I’m at Reinforcement Learning Conference in Montreal this week! 🇨🇦 Come to our Model-Based RL workshop tomorrow! Our speakers are cool!!
Chenhao Li@breadli428Apr 22
📢 Call for Papers! #RLC2026 🇨🇦 🌎 We are now inviting contributions to the Workshop on Model-based RL in the Era of Generative World Models at @RL_Conference in Montreal, Canada! 🇨🇦 🔗 Webpage worldmodels-rlc.github.io/ 📄 Submit paper now! openreview.net/group?id=rl-c… 🧵Format
3
1
52
14,290
Load more