Théo Vincent and I recently dived into BBF, a fascinating algorithm that is still poorly understood. Feel free to checkout the results of our analysis:

I don't disagree that data will fix many current limitations, but unless Figure is willing to visit people's apartments to periodically collect data for their evolving use cases, we won't see a PMF. The scalable fix is to learn in deployment, which requires new algorithms.

How do robots thrive in dynamic worlds? Join us at IROS LeapRiDE 2025 Full-day Workshop Oct 20, ROOM 102A, Hangzhou HIEC, China 🎯 From quadruped to humanoid ping-pong & robot soccer 🎓 Speakers from MIT, Berkeley, DeepMind, NTU, NUS, Monash, Tongji, & KU Leuven #IROS25

Our Reinforcement Learning Group is looking forward to a hearing from Calarina Muslimani next week on October 23rd as she presents "Towards Improving Reward Design in RL: A Reward Alignment Metric for RL Practitioners" Thanks to Rahul and Gusti T. Winata for organizing

I'm presenting four different works at IROS 2025 this week in Hangzhou 🤖

🔊Simplicial Embeddings (SEMs) Improve Sample Efficiency in Actor-Critic Agents🔊 In our recent preprint we demonstrate that the use of well-structured representations (SEMs) can dramatically improve sample efficiency in RL agents. 1/X

1/9 Can we leverage foundation models for better RL agents? Yes! We introduce Language-Aligned Reward Machines (LARMs) and a framework that uses Foundation Models (FMs) to automatically generate them. This work enables new ways to train RL agents efficiently on complex tasks 🧵

Very happy to share that "iterated Q-Network (i-QN)" received a J2C Certification from TMLR🥳 What is i-QN about?👇 x.com/Theo_Vincent_/…

New #J2CCertification: Iterated $Q$-Network: Beyond One-Step Bellman Updates in Deep Reinforcement Learning Théo Vincent, Daniel Palenicek, Boris Belousov, Jan Peters, Carlo D'Eramo openreview.net/forum?id=Lt2H8… #reinforcement #iterative #iterations

I am sharing this post with a heart full of gratitude. I am deeply humbled and honored to be named a recipient of this year's Alfried Krupp Förderpreis. Thank you to the Krupp Stiftung for this prestigious award and for a truly memorable and beautifully organized ceremony. This

The Gran Turismo research that I've worked on at Sony AI just got accepted at #AAAI as an oral! We'll publish it on arXiv soon. Looking forward to see you in Singapore!

The SPARC paper is online! (link below) During my internship at Sony AI, we created a policy that can generalize across all cars in Gran Turismo 7. Even to unseen cars, without knowing any vehicle details! Co-authors: Patrick MacAlpine, Kaushik Subramanian, Peter Stone, Peter Wurman

We are very happy to welcome Jan Peters at GRASP Laboratory , UPenn and give a talk on “Inductive Biases for Robot Learning.” After talk, students like me were fortunate to have lunch together. To be honest, at first, I expected heavy math/physics priors, but his message was