Learning Robot Learning Hackathon

​Robot learning is on track to be the defining technology of our lifetime, but unlike LLMs or agents, you can't just start with a laptop. You need an arm, and for many models, a strong local GPU.

​This request keeps coming up at our paper clubs: engineers and computer scientists who want to try things hands-on but don't have an arm. We have a lot of big arms and thanks to the Pollefeys lab, we'll have over a dozen SO-100 arms for the weekend.

​Format
Open hacking. No prizes, no judging, no demo pressure - the point is to learn and to meet other people with shared interests.

​What we provide

• ​SO-100 arms and TRLC-DK1/YAM arms for those with prior experience.

• ​An intro session giving you a brief overview of robot learning, recommendations for starting points and tips on what to avoid.

• ​Food, drinks and snacks for the weekend.

​What you need

• ​Comfort reading and writing code. Ideally you already have ML experience and just never had the time to point it at robotics.

• ​Your own laptop.

• ​Initiative: we'll help you figure things out, but we won't teach you.

• ​Optionally, a friend or two to work together with, so we can make the most use of the arms per person.

• ​You should be ok spending <50$ on compute to train a few models. We have four 5090s at the office, but we'll be using them for inference.