CLARE has recently been accepted to RA-L! We propose a lightweight continual learning method for VLAs that injects modular adapters into selected VLA modules and autonomously expands the model only where necessary. Paper: arxiv.org/abs/2601.09512 Website: tum-lsy.github.io/clare

Ken Glodberg’s talk happening now at the Workshop on Semantics for Reliable Robot Autonomy: From Environment Understanding and Reasoning to Safe Interaction at VIP lounge C #ICRA 2026 – at Vienna, Austria

Our work, "A Primer on SO(3) Action Representations in Deep Reinforcement Learning," was accepted to #ICLR2026! We provide a systematic study of action representation choices in RL, showing that they fundamentally impact training stability and performance. #Robotics #AI #RL

From Demonstrations to Safe Deployment: Path-Consistent Safety Filtering for Diffusion Policies Paper: arxiv.org/abs/2511.06385 Website: tum-lsy.github.io/pacs/

SM2ITH: Safe Mobile Manipulation with Interactive Human Prediction via Task-Hierarchical Bilevel Model Predictive Control Paper: arxiv.org/abs/2511.17798 Video: youtube.com/watch?v=ld-Iyxqk…

ICYMI: We had a blast at #AAAI 2026 in Singapore last weekend presenting our @IeeeTro paper #SICNav: Safe and Interactive Crowd Navigation Using Bilevel MPC and Bilevel Optimization. Paper, Code, Videos: sepehr.fyi/projects/sicnav

Happy Robotic Holidays from the LSY Lab! 🤖🎄 Simulation + Domain Randomization = Robust Policies. In this video, we randomize: 📸Camera 🖼️Background 💡Lighting Excited about Sim-to-Real? Come work with us!🚀 Created by Timo Class @OliHausdorfer #Robotics #Sim2Real #AI

How do you teach robots to make good decisions in the real world? 🤖 We have developed a course series that bridges optimal control, MPC, and RL—down to hands-on robot implementations. Course materials are coming soon. Subscribe here for updates: groups.google.com/g/safe-rob…

We are excited to share our recent collaboration with @Bitcraze_se, showcasing how Bitcraze’s new Crazyflie Colour LED Deck adds a new dimension of expression to our SwarmGPT system for interactive drone choreography design. Hope you enjoy the video and happy holidays!

“Let’s start with a bang — big energy, right away.” With SwarmGPT, anyone can describe the feeling of a performance in plain words. Paper: ieeexplore.ieee.org/abstract… Videos: youtube.com/playlist?list=PL… Website: utiasdsl.github.io/swarm_GPT… #SwarmRobotics #GenerativeAI #TechMeetsArt

Diffusion Policies and VLAs with Formal Safety Guarantees🛡️ We just released our new paper: “From Demonstrations to Safe Deployment: Path-Consistent Safety Filtering for Diffusion Policies” 📑Paper: arxiv.org/abs/2511.06385 🌐Website: tum-lsy.github.io/pacs

Our paper “Failure Prediction at Runtime for Generative Robot Policies” has been accepted to #NeurIPS 2025! 📄 arxiv.org/abs/2510.09459 🌐 tum-lsy.github.io/fiper_webs… 💻 github.com/utiasDSL/fiper #Robotics #MachineLearning #AI #GenerativeAI #Safety

🚀Deploy VLAs & diffusion policies on real robots with 𝗖𝗥𝗜𝗦𝗣 – compliant ROS2 controllers for learning-based manipulation & teleoperation! Python, Gym and LeRobot interfaces, demo-ready. Paper & code: github.com/utiasDSL/crisp_co… #robotics #VLAs

Prof. Schoellig is giving a virtual seminar tomorrow, Aug. 26 (Tue) @ 13:00-14:30 (CEST) / 7:00-8:30 (EST) on "Aerial Swarms: From Safe Motion Planning to Language-Based Interaction." Live stream: youtube.com/live/RtRyaqodld0… @TU_Muenchen @TUM_MIRMI @UofT @UofTRobotics

The story about our crowd navigating robot aired this morning on the German public broadcaster ZDF! ZDF video: zdf.de/nachrichten/zdf-morge… w/ subtitles: youtu.be/Sa_pYF-RNsk

We hosted @ZDF @morgenmagazin to film a story about our crowd navigation project running SICNav. It should air on Monday 12 May 2025 between 05:30 and 09:00. SICNav @IeeeTro paper: sepehr.fyi/projects/sicnav/ Find us at #ICRA2025: WeET14.6, Wed. 21 May 17:05, Room 402.

Are you a TUM student with a passion for robotics, machine learning and control? Check out our course offerings for the upcoming summer semester here: ce.cit.tum.de/lsy/teaching/