This is how Microduck learns to walk. Reinforcement Learning, explained by ducks. 🦆 (impeccable) music arrangement by Antoine Pirrone

@antoinepirrone This is how I’m convincing my niece to join 4H kids robotics group. Thank you! 🙏

@antoinepirrone Ok my post is already worth it, thank you

@antoinepirrone Reachy mini can do the same with an integrated 360-degree rotation system

@antoinepirrone this is epic

@antoinepirrone I watch it on repeat just for the music at the end

@antoinepirrone there's definitely a crescendo to the whole thing

@antoinepirrone I bought a duck on pure impulse and this video just taught me a lot. Thanks!

@antoinepirrone What are your rewards ? Are u implementing gait + rl rewards ?

@antoinepirrone Check the repo:

@antoinepirrone I love this soooo much. So good 🦆

@antoinepirrone This is awesome

@antoinepirrone the best explainer for reinforcement learning i’ve ever seen

@antoinepirrone Naruto also taught this technique. 😉 During his leaf splitting exercise.

@antoinepirrone I almost made the duck do the signs with its feet

As a trendy toy KOL, I’ve been invited to attend the China Toy Expo next month. While there will be exhibitors from all over the world, I doubt I’ll see *this* duck there—it represents a truly AI-powered toy. What I expect to see instead are likely just plush dolls stuffed with large language models.😼

@antoinepirrone Haha, great video

@antoinepirrone Is this how those chinese olympiade robots are trained?

@antoinepirrone I actually cannot wait for my lavender microduck😄 sooo good

@antoinepirrone Watching a 30 cm duck drive 14 motors in sim is oddly satisfying. Sim-to-real is the final race 🦆

@antoinepirrone The way the reward shaping forces each fall and reset really nails the trial and error loop, looks like a solid RL demo

@antoinepirrone Yeah. Awesome. March on ducks. 😀

@antoinepirrone Love the video hahaha

@antoinepirrone 🔥

@antoinepirrone This is a great concept for teaching.

@antoinepirrone It reminds me so much of the @OpenAI incident at @huggingface . Communication and sharing experiment results enabling agents to improve on a goal!

@antoinepirrone This is incredible! I'd love to see more of these.

@antoinepirrone The penguin movement is actually way smoother than I expected ngl

@antoinepirrone Dude, such a crazy and fun explanation !! 😍

@antoinepirrone how the duck knew what the walking is if anyone not show them what walking is and how it look like, and decide we learn walking they try random move but how they figure this move are wrong and this are right and we now going in right direction

@antoinepirrone joust accessory and game mode would be a nice game for micro duck.

@antoinepirrone Are the Microducks going to have a set of Heelys they can zip around on?

@antoinepirrone @MicroDuckNVDA see that