Bram Grooten

My PhD thesis is finished! It's all about making reinforcement learning agents more adaptive to changes in their environment. Here's the audio summary by NotebookLM. See the full pdf on my website (link below)

@zevML In multiple ways! See the 1-page summary on page 11 of the pdf: bramgrooten.nl/thesis

@BramGrooten Adaptive RL is the production gap. Models trained on static environments fail when reality shifts. How did you approach distribution shift in your work?

My PhD thesis is finished! It's all about making reinforcement learning agents more adaptive to changes in their environment. Here's the audio summary by NotebookLM. See the full pdf on my website (link below)

@jonas_kg @adaption_ai Good point, thanks for tagging them. I hope @sarahookr likes the thesis!

@BramGrooten @adaption_ai

Full pdf: bramgrooten.nl/thesis I'll be defending this work on April 8th, 16:00 at TU Eindhoven. Let me know if you'd like to be there, or if you'd like a physical copy of the book 🤓

🔥 Job alert 🔥 We have two exceptional vacancies for research engineers to work on OpenML. Join us to work on open source software full-time in an exciting research lab and make AI more accessible and reproducible.  Apply now: amore-labs.github.io/website/join/j… Please share! ❤️❤️❤️

Join the international boycott now.

I've written a quick blog post on VLA papers that were presented at AAAI a few weeks ago: bramgrooten.nl/blog/vlas-at-a…

We organized an RL competition during the first Openmind Research Institute Winter School in Malaysia. The participants were able to implement SARSA and SAC in just 2 days onboard our Embodied MuJoCo Ant! 🎉

I should probably read the UniVLG paper, heard it multiple times here at AAAI

When multiple people mention a paper to you, you know you should probably read it

Staying calm while AGI sparks are flying around. "This is what it feels like to do AI research nowadays" --Yolanda Gil, in her invited talk at AAAI (she showed this picture of Nikola Tesla reading a book in his lab)

Well done, @s_albrecht 👏🏻

The most important session of #AAAI is overflowing

I'm at #AAAI in Singapore to present our SPARC paper. Hope to see you today! 12pm: poster 1356 in Hall 4 3pm: oral in the RL 3 session, room Garnet 219

@robot_in_space2 And thanks for the interesting book tips in your blog!

@robot_in_space2 On all the other predictions I agree though. That those are things we'd like to see :)

Here are 8 forecasts for the next 25 years (faster travel, space, AI, new governance, more 👀) 👉lupusorina.github.io/blog/2026/fore… Enjoy. 🌎🚀🤖 After all, this is the future I'd like to help build.

Awesome 4-day RL winter school coming up right after AAAI, close to Singapore. You can still try to apply, you never know!

FWIW, my New Year's resolution a few years ago was to use metric when discussing distances, temperatures, etc., and I'm still sticking to it. If you're looking for a good New Year's resolution this year, I figure we just need ~340 million more to join and we're there!

Our paper was accepted as an oral at AAAI 2026. Hope to see you there!

And on Wind-MuJoCo environments, SPARC also shows the best robustness to unseen wind speeds.

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, @kausubbu, @PeterStone_TX, Peter Wurman