Carlo D'Eramo

Using only box-forwarding speed as the reward, our Stackelberg PPO automatically evolves robots with arms for pushing and legs for moving. The key idea is a novel game-theoretic view of structure–control co-design, yielding more effective optimization and dramatically better designs. Come see our poster at ICLR 2026 on Apr 25, 10:30 AM, at P4-#4810. With @YuhuiWangAI, @YanningD_AI, @oneDylanAshley. Paper: arxiv.org/abs/2603.15388 Project Page: yanningdai.github.io/stackelberg-pp…

🧠 What we are looking for: • Rigorous ML/RL background & math fluency • Excellent programming skills • Analytic problem-solving & communication If you want to push the boundaries of RL, join us! Please RT to help spread the word! #ML #RL #AI #PhDPosition

📢 We are seeking a highly motivated student for a 3-year fully funded PhD position in Reinforcement Learning at the University of Innsbruck! 🇦🇹 Help us advance the theory & algorithms of off-policy RL. Details & Apply: samueletosatto.online/new-phd-positi…

🧵 Accepted at @iclr_conf ! Target networks stabilize bootstrapping in RL 🛡️ But induce slow-moving targets 🐢 Online networks adapt fast ⚡ But can diverge with function approximation 💥 𝗠𝗜𝗡𝗧𝗢🌿 uses the online network 𝗼𝗻𝗹𝘆 𝗶𝗳 𝗶𝘁 𝗰𝗮𝗻 — yielding faster and more stable RL. Here’s how 👇

It was a pleasure to collaborate with @YogeshTrip7354, Tim Faust, @aportekilaa, Yaniv Oren, Melih Kandemir, @Jan_R_Peters, and @CarloDeramo Thanks to @ias_tudarmstadt, @DFKI @Hessian_AI, @infsys_uniwue for supporting this research!

🌿 MINTO has been accepted at #ICLR2026! 📌 MINTO is a simple, yet effective target bootstrapping method for off-policy RL that enables faster, more stable learning and consistently improves performance across algorithms and benchmarks. 📄 Preprint: arxiv.org/abs/2510.02590

I had a fantastic time discussing my research @AmiiThinks and @UAlberta last August. If you are interested in Multi-Task Reinforcement Learning (MTRL) and Mixture of Experts (MoE), then this talk is for you. ​➡️ Full talk: youtu.be/yqRajcJLl2I #reinforcementlearning #AI

In this paper, the authors compute the gradient update of the policy of one agent by accounting also for the update of all other agents. I feel this is a fairly general idea that could be applied to most multi-agent RL algorithms. 🔗arxiv.org/abs/2509.12117

Attending @RL_Conference this year at @UAlberta, Canada 🇨🇦 ! I'm excited to be organizing a workshop on Inductive Biases in Reinforcement Learning @ibrlworkshop. Let's meet to discuss multi-task RL, Mixture of Experts, and other cool topics. Hope to see you there!

RLC starts tomorrow here in Edmonton. I couldn't be more excited! It has a fantastic roll of speakers, great papers, and workshops. And this time, it is in Edmonton 😁 @RL_Conference is my favourite conference, and no, it is not because I am one of its organizers this year.

Looking forward to @RL_Conference! I will be presenting 4 posters, feel free to come and exchange with me during the conference, @RLFrameWorkshop, or @ibrlworkshop🙂

🗓️ The IBRL Workshop kicks off tomorrow! 🎉 Join us at @RL_Conference @UAlberta to explore how Inductive Biases can boost 🚀 the performance of RL agents. 📄 Accepted papers: sites.google.com/view/ibrl-work… 📅 Full schedule: sites.google.com/view/ibrl-work… #ReinforcementLearning #RLC2025

To increase the reward propagation in value-based RL algorithms, it is tempting to reduce the target update period🤔 But, this makes the training unstable💔 @RL_Conference, I will present i-QN, a new method that allows faster reward propagation, while keeping stability⚡️ 👉🧵

🎤 Very excited to give a talk @Cohere_Labs next week 🎤 I will be presenting the research I have been working on for the last 2 years with @CarloDeramo, @Jan_R_Peters, and many more collaborators! x.com/Cohere_Labs/st…

Thanks to my amazing co-authors: Tim Faust, @YogeshTrip7354, @Jan_R_Peters, and @CarloDeramo 🎬Video👉youtu.be/xUpvOTyetKY 📄Paper👉arxiv.org/pdf/2503.01437 👨‍💻Code👉github.com/theovincent/Ea… 🏋️‍♂️Weights👉huggingface.co/TheoVincent/At…

📢 Submission deadline extension: the new deadline is June 6th AoE 🔗 Portal: openreview.net/group?id=rl-co… 🌐 More information at: sites.google.com/view/ibrl-work… 🚀 Looking forward to seeing you at @RL_Conference !

🚀 TU Darmstadt leads the new Cluster of Excellence "Reasonable AI" – advancing trustworthy, efficient & adaptive AI grounded in common sense. A big thank you to the entire team, the university, and the state of Hesse for their tremendous work & support! buff.ly/N7xLKfW

I'm pleased to announce that, starting October 1st, I will be joining the Computer Science department at Lund University (in Sweden) as a Senior Lecturer. I will join the Robotic and Semantic Systems group and collaborate in the RobotLab LTH, where I'll bring my RL expertise.

I'm very excited for this new adventure and very grateful to @Jan_R_Peters for his massive support during my time as postdoc/group leader at @ias_tudarmstadt Also, many thanks to my amazing friends @GeorgiaChal and @CarloDeramo for their help and support.

How do you tune the hyperparameters of your RL agent? 🤔 Come and chat with me about it tomorrow afternoon at poster 397 of @iclr_conf ! I will be presenting⚡️ Adaptive Q-Network⚡️a method that adaptively selects the hyperparameters of your RL agent during training 🔥