Marc Raibert: Boston Dynamics | MIT Artificial Intelligence (AI) Podcast

Share it with your friends Like

Thanks! Share it with your friends!

Close

This is a talk by Marc Raibert for course 6.S099: Artificial General Intelligence. He is the CEO of Boston Dynamics. This class is free and open to everyone. Our goal is to take an engineering approach to exploring possible paths toward building human-level intelligence for a better world.

Note: Due to technical difficulties, we don’t have a screencast of the slides, and the video of the slides is low resolution. Despite this, I chose to include several parts of the talk that show slides, especially with videos. It’s not optimal, but I hope you learn and enjoy anyway. Thanks for understanding. We’re always learning and improving.

Note 2: Marc and I asked people not to release the videos around the 20 minute mark before they are officially released by Boston Dynamics. They have been now, go check them out: https://www.youtube.com/user/BostonDynamics

OUTLINE:
0:00 – Introduction
1:06 – Slides
24:27 – Demo
33:40 – Q&A

INFO:
Course website: https://agi.mit.edu
Contact: agi@mit.edu
Playlist: http://bit.ly/2EcbaKf

CONNECT:
– If you enjoyed this video, please subscribe to this channel.
– AI Podcast: https://lexfridman.com/ai/
– Show your support: https://www.patreon.com/lexfridman
– LinkedIn: https://www.linkedin.com/in/lexfridman
– Twitter: https://twitter.com/lexfridman
– Facebook: https://www.facebook.com/lexfridman
– Instagram: https://www.instagram.com/lexfridman
– Slack: https://deep-mit-slack.herokuapp.com

Comments

Czypu says:

boston dynamics + ASI = bye humanity

Adam Roder says:

So why start with quadriped/biped motion? You're fighting natural evolution. Snakes came earlier and still survive.

Chyrre says:

Insurance company:

New guy at Boston Dynamics: Robotic mule smashed my car

Gustavo Bakker says:

Didn't know that Michael Keaton was into AI

Jameel Ja says:

Is there any research on synthetic muscle fibers?

Dae-Suk Kim says:

I think that Boston dynamics needs to test stability tests using other ways not by humans. It looks not good although they are robots…

Subhajit Sahu says:

can the spot mini robot think it would be easier to go through the door if it scared andy?

Tom Harrison says:

"We're thinking about making the Spot Mini a platform like Android, for robots." — androids are robots, right? Brilliant and inspiring presentation.

MultiGladiator says:

impressive talk. this gives me motivations. thanks for your efforts again

Jason Voss says:

I love the tech. But the potential use by nasty pasties with malicious intent is a bit concerning. High time we realized we are all in the same tribe.

Jon Dough says:

52:46 My favorite part

John theux says:

38:40 Just add skin on it so it will be less harmful.

miraculixx says:

I wonder if in his opinion these machines are intelligent?

John R says:

That was awesome thx for sharing!

Essam Al-Mansouri says:

50:50
"How did you make the robots really fast?"
"How did we make them fast?"
"No, my question is, how did you make them, fast?"
I'm actually impressed that the kid understood that Raibert misunderstood the question, and clarified it without changing a word.

Cashel McKenzie says:

wow he made sure to answer the kids questions fully without patronizing, what a legend

John Batchler says:

Did you saw the robotics that I'd build at campus??

rubel hasan says:

thank for this video upload

alexinaboxx says:

Would any product made be considered an import for the US economy vs an export now that they are owned by a Japanese holding company? Would America have to import its own innovation?

Zheng Cheng says:

Thanks for upload this video but why top left corner slider so muggy? it's like 240p slides

Pierre-Emmanuel Féga says:

Thank You Lex for enlightening a disruptive field that is still lying behind the shadows of fear and incertitude. All the best

MrChatmoon says:

thank you for sharing.

MrChatmoon says:

thank you for sharing. amazing piece of work.

How Does it Really Work says:

Highlights:

8:22 "Make low levels very robust to disturbances, so that the planning steps do not have to take care of the minutiae of the real world"

9:55 "Treat the control system + robot hardware + the environment holistically"

24:26 Spot Mini demo

38:42 "Safety is a major unsolved problem"

48:12 Presently the robot does not use learning — instead its designers make very simple decisions on how to divide the state space and apply different controllers (also 7:13) On top of this, there is an ad hoc application for driving robot for specific tasks (49:24)

Taylor Caforio says:

Cant make all of these, so thanks

Jason Gooden says:

They should put fur on Spot Mini….

eerereps says:

Very informative!

Ragecraft says:

Knowledge-workers will be the first to be replaced by AGI because their bodies are irrelevant. Replacing laborers with AGI makes less sense from a financial perspective because human laborers are innexpensive relative to knowledge-workers while android bodies are expensive and will require a similar investment in processor hardware.

Write a comment

*