@leggedrobotics

The Robotic Systems Lab designs machines, creates actuation principles, and builds up control technologies for autonomous operation in challenging environments.

 Zurich, Switzerland
rsl.ethz.ch
Joined March 2016
  • Tweets714
  • Following171
  • Followers39K
368Media
This work is led by Mohamad Danesh @mo_danesh, together with Chenhao Li @breadli428, Amin Abyaneh, Anas Houssaini, Kirsty Ellis, Glen Berseth @GlenBerseth, Marco Hutter, and Hsiu-Chin Lin. @mcgillu @Mila_Quebec @UMontreal @ETH_en @ETH_AI_Center @leggedrobotics
1
2
807
We will see you in Austin, Texas, US, in November at CoRL 2026 @corl_conf!
1
522
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
56
4,765
LP-ACRL removes the need to predefine task difficulty. It measures task-wise learning progress from episodic-reward changes, then uses a softmax to shift sampling toward the current learning frontier, where tasks show stronger positive competence gains at that stage.
1
1
437
more replies
After teacher–student distillation, a single policy transfers to ANYmal D and achieves: • 3.0 m/s on flat terrain • 2.5 m/s on challenging terrain • Up to 3.0 rad/s angular velocity across all tested terrains
1
1
430
Scaling Rough Terrain Locomotion With Automatic Curriculum Reinforcement Learning Authors: Ziming Li @ZimingLi97, Chenhao Li @breadli428, Marco Hutter Published in: IEEE Robotics and Automation Letters ( Volume: 11, Issue: 8, August 2026) ieeexplore.ieee.org/document…
382
We achieved high-speed rough-terrain locomotion on ANYmal with an automatic curriculum. LP-ACRL automatically samples terrain types, levels, and velocity commands at the correct time based on policy performance, without predefining an order. 🔗sites.google.com/view/lp-acr… 🧵RAL 2026
5
26
4
178
19,148
Real-world locomotion does not follow a single easy-to-hard axis. Terrain type and geometry compound with linear and angular velocity specifications. In our scaled setup, their Cartesian product forms 600 tasks, making a manually ordered curriculum impossible.
1
880
VLMs don't understand a robot's body. How can a robot learn from its failures without retraining? PragmaBot: online in-context learning from real-world experience. The robot reflects on failures, stores lessons in memory, and retrieves them for new tasks. RA-L · #IROS2026
2
7
1
54
4,854
more replies
How much memory belongs in the context? 5 random memories: 17% first-action accuracy on unseen tasks. Entire memory in the prompt: 74%, at 7.5x the tokens. Top-k relevant via RAG: 89%. Relevant context beats more context. As memory grows, what should a robot forget?
1
1
352
Congrats to Kaixian Qu, Guowei Lan, René Zurbrügg, Changan Chen, Christopher Mower, Haitham Bou-Ammar & Marco Hutter. Say hi at IROS 2026! ieeexplore.ieee.org/document… arxiv.org/abs/2507.16713 pragmabot.github.io github.com/leggedrobotics/pr…
1
310
We are releasing EgoHTR, a dataset with both human motions and terrain references, accepted @corl_conf. 📖 Paper: lnkd.in/eeepUefs 🌐 Project Page: egohtr.github.io • 55 scene-aligned sequences • 150k+ frames • rough-terrain environments • multi-modal 3D scene
3
9
2
43
11,892
We’re open-sourcing the dataset and the full data pipeline to make this resource accessible to the community and hopefully enable new directions in learning-based robotics.
1
1
1,059
By Alex Brandes, Haig Conti Georges Sajelian, Manthan Patel @patelm99, Dominik Hollidt, Chenhao Li @breadli428, Matthias Heyrman, Oliver Hausdörfer, Manuel Kaufmann, Xi Wang, Jonas Frey, Angela Schoellig @angelaschoellig, Christian Holz, Marc Pollefeys @mapo1, and Marco Hutter.
2
586
PPO has long dominated robot locomotion training in simulation. SAC, despite its sample efficiency, couldn't keep up. We analyze why: 🔗sabagian.github.io/sac_relea… 🔥Integrated into RSL-RL, our approach requires only minimal changes, making SAC a drop-in alternative out of the box.
7
42
5
342
45,877
more replies
This project focuses on understanding the performance gap between PPO and SAC in massively parallel robot learning. While efforts like FlashSAC @hojoon_ai and FastSAC @younggyoseo explore separate development, RSL-RL-SAC is made to stay close to the widely used RSL-RL codebase.
1
5
1,606
This project is led by Gianluca Sabatini, supported by Chenhao Li @breadli428 and Marco Hutter @leggedrobotics. We thank Clemens Schwarke's implementation insights. Check out the paper! arxiv.org/abs/2605.24975

Bridging the Gap: Enabling Soft Actor Critic for High Performance...

Proximal Policy Optimization (PPO) has become the de facto standard for training legged robots, thanks to its robustness and scalability in massively parallel simulation environments like...

arxiv.org
8
1,194
Load more