Combining real-time interactivity, task understanding, and full-body action prediction on a humanoid is so, so hard.
Here's an example where we bring all of these together in Gemini Robotics 2 🤖🧠
Excited to share Gemini Robotics 2!
6 months ago, I wrapped up my PhD at Stanford to join GDMR full-time.
Since then, I’ve learned so much from the team, while working on video understanding for our ER agent and improving the robustness & instruction-following of our action
One brain. For any robot. 🤖
We’re launching Gemini Robotics 2: our next-generation physical AI bringing full body intelligence to humanoids, advanced dexterity, multi-robot teamwork and more.
New work led by @riadoshi21 on training a single VLA policy for multi-robot collaboration
Excited about all the new kinds of tasks this can unlock, when robots can coordinate and work together!
🤔 Can we train one VLA policy to control multi-robot teams without any explicit communication?
✨ Introducing CHORUS: a single policy for decentralized, multi-embodiment collaboration
🧵⬇️
And it's even better in-person! Got to see Memo live a few weeks ago and it's such a great design :) Love the gloves, seems to enable a scalable path to high quality data. Huge congrats to the team, especially @tonyzzhao so impressed by your resilience over the past 2 years!
Today, we present a step-change in robotic AI @sundayrobotics.
Introducing ACT-1: A frontier robot foundation model trained on zero robot data.
- Ultra long-horizon tasks
- Zero-shot generalization
- Advanced dexterity
🧵->
How should an RL agent leverage expert data to improve sample efficiency?
Imitation losses can overly constrain an RL policy.
In RL via Implicit Imitation Guidance, we show how to use expert data to guide more efficient *exploration*, avoiding pitfalls of imitation-augmented RL
