Mohamad H. Danesh (@mo

Mohamad H. Danesh

613 posts

Mohamad H. Danesh

@mo_danesh

phd @McGillU and @mila_quebec, working on rl and robot learning / ex- @LetsUnifyAI, @NUSComputing, @EngineeringOSU

Montréal, Québec

Joined February 2017

Pinned
Mohamad H. Danesh
@mo_danesh
May 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.
GIF
GIF
Mohamad H. Danesh
@mo_danesh
Jul 26
For NeurIPS rebuttals, can we respond to the AC's meta-review? We can't find a way to do so on OpenReview and there is nothing about it in the handbook. #NeurIPS
Mohamad H. Danesh
@mo_danesh
Jul 1
Angine de Poitrine?
Mark Carney
@MarkJCarney
Jul 1
The Eurovision Song Contest just got a lot tougher. Because next year, Canada is in.
Mohamad H. Danesh
@mo_danesh
Jun 15
What should a robot dream about? 🤖💭 IMO the debate over whether robots need world models is over. Now, the real fight is over "what kind". In this blog post, I break down the 5 major "fault lines" dividing the field of physical AI today. Read my notes here 👇
modanesh.github.io
What Should a Robot Dream About?
Notes on world models for robotics.
Mohamad H. Danesh
@mo_danesh
Jun 9
Our method adds as little as ~10 lines on top of TD3+BC: pull generated actions toward the data, push them apart from each other. Go and check it out!
Mohamad H. Danesh
@mo_danesh
Jun 5
The code is now available! 🚀 DriftQL learns a one-step Q-guided actor corrected by a learned drift field. We beat baselines with no denoising, solvers, auxiliary actors, or distillation. 💻 Code: github.com/anashoussaini/…