Robotics will Evolve with Generative AI

Google DeepMind robotics papers, but something even more surprising out of Stanford.

Hey Everyone,

As we head into 2024, it’s not just Generative A.I. that is in a secular research boom, it’s the field of robotics as well.

What’s interesting to me is a lot of this is also occurring under the radar of most media coverage. As stunning as advancements were in 2023 with Generative A.I., don’t sleep on robotics going more mainstream in the next few years.

This post by

Conrad Gray

of Humanity Redefined Newsletter.

A few days ago Google DeepMind, launched three new papers for advanced robotics that show how we can help robots make faster decisions, and better understand and navigate their environments.

• The first, AutoRT, brings together large vision and language models with robotic control models in a single system, which could help create robots that understand practical human goals in much more natural ways. The resulting system safely orchestrated as many as 20 robots simultaneously, gathering a diverse dataset across 6,650 unique tasks.

• The other two papers present a new system for scaling up the efficiency and speed of Robotic Transformers (SARA-RT), and a system for improving the generalization of learning robotic motions by sketching trajectories over top training videos (RT-Trajectory).

Learn more

• AutoRT paper

• AutoRT GitHub

• SARA-RT paper

• SARA-RT website

• RT-Trajectory paper

• RT-Trajectory GitHub

• Access the Open X-Embodiment dataset

Here’s the thing though, GDM in collaboration with Stanford’s mobile ALOHA robot also showed off the prowess of how fast tech has moved along.

Introducing Mobile ALHOA

I really like their videos, I’ve taken the privilege of embedding a few in this post.

Stanford’s mobile ALOHA robot