Multi-agent robot learning: if that's your thing, you might be interested in our workshop at IROS 2023. djhanove.github.io/IROS23_MR… We have some great speakers, and if you have work to share, we are looking for submissions by August 20th!
These robots were told one thing: score. They taught themselves to play soccer – creating their own strategies and their own ways to walk, dribble and block. cbsn.ws/43GDktB
Football players can tackle, get up, kick and chase a ball in one seamless motion. How could robots master these motor skills? ⚽
We trained AI agents to demonstrate these agile behaviours using end-to-end reinforcement learning.
Find out more: dpmd.ai/41N3CsM
I have an unusual story to tell about how score-based generative models construct the data manifold. Will present this at @NeurIPS 2022 Score-Based Model workshop. A really enjoyable collaboration with @benm and many at @DeepMind
paper: openreview.net/forum?id=eSZq…
You might think it’s just a video, but you’d be mistaken. We just released a series of “explorable videos” wherein @3blue1brown teaches you about quaternions and 3d rotation! Is this a new educational medium? eater.net/quaternions
We would ask you to guess what this is, but it's just not going to happen. It's the paths of 800 unmanned bikes being pushed until falling over, from Matthew Cook.
instagram.com/p/BkZh2ZQH1eB/… ^JH
This looks handy: "great little utility from our team for making it easier to work with jupyter notebooks in code repositories github.com/ttung/nbencdec (via @thefreemanlab)"