MIT’s Russ Tedrake Says Robotics Is Finally on a Rocket Ship

Physical AI is having a moment.

But Russ Tedrake says this robotics boom may be different from the ones that came before.

In this episode of Automated, Brian Heater speaks with Russ Tedrake, Toyota Professor at MIT and founder of a stealth physical AI startup, about why robotics may finally have the talent, investment, manufacturing capacity, and AI breakthroughs needed to reach the next stage.

They discuss why humanoid robots have become surprisingly capable, why machine learning has moved ahead of theory, and why robotics engineers now have to study the systems they build in a very different way.
Brian and Russ also dig into large behavior models, vision-language-action models, robot data, simulation, deployment, and why the next milestone for physical AI is getting more robots into the real world.

Finally, Russ explains why he is thinking deeply about the future of work, and why the most thoughtful robotics companies should focus on amplifying people rather than replacing them.

If you want a clear, grounded look at why physical AI may finally be entering a new era, this is the conversation.

KEY MOMENTS
(00:00) Why machine learning is ahead of theory
(02:16) Russ Tedrake’s Detroit roots
(02:42) The Ford plant internship that taught him a hard automation lesson
(05:38) How Russ got into bipeds and MIT’s Leg Lab
(07:04) Passive dynamic walkers and learning to walk
(08:33) Why humanoid robots have become surprisingly turnkey
(10:48) Early reinforcement learning in robotics
(12:18) Why simulation changed robot locomotion
(13:46) Why AI has outrun theory
(16:08) The fundamental questions behind robot learning
(18:20) What it is like to be a student in AI right now
(19:37) How to avoid chasing every shiny AI object
(22:51) Why this robotics moment feels different
(25:31) Why Boston is a powerful robotics ecosystem
(26:39) Drake and model-based robotics software
(27:42) Large behavior models at TRI
(28:38) The difference between LBMs and VLAs
(30:54) Why the robotics data problem is often framed wrong
(34:34) How different data sources connect
(35:25) Edge cases, robustness, and multitask pre-training
(36:32) Why deployment is the next major robotics milestone
(38:01) Why Russ believes physical AI may have escape velocity
(40:07) Why “amplifying, not replacing people” matters
(43:08) What empathy means in physical AI
(45:13) Why Russ values running and biking to work

Connect with Russ Tedrake
https://www.linkedin.com/in/russ-tedrake-88648a4a

Learn more about Russ Tedrake at MIT
https://locomotion.csail.mit.edu/russt.html

Learn more about Drake
https://drake.mit.edu/

Learn more about Large Behavior Models from Toyota Research Institute
https://toyotaresearchinstitute.github.io/lbm1/

We’d love to hear from you. Have thoughts or guest suggestions?
Reach us at podcast@automate.org

You can find the transcript and more episodes of Automated at automated.fm

Unlock full access to Automated and explore everything automation.
Subscribe today and leave a review on YouTube, Apple Podcasts, and Spotify. Also, subscribe to the Automated Newsletter.
https://www.youtube.com/@automatedpodcast
https://podcasts.apple.com/us/podcast/automated-with-brian-heater/id1837762221
https://open.spotify.com/show/60olq6brlBEIJWggx2fMR6
https://www.automate.org/automation/automated-newsletter

You can also find us on:
LinkedIn https://www.linkedin.com/showcase/automated-podcast-by-a3/
Instagram https://www.instagram.com/automatedpod/