Extreme parkour with legged robots • Hacker News Board

Re: Extreme parkour with legged robots

#191

Post by danmaz74 » Wed, Oct 11, 2023, 9:20 AM UTC

waldarbeiter wrote: post #159

danmaz74 wrote: post #137

I scanned the paper and, if I got it right, during training the robot gets "external" information about its world position and speed compared to pre-set waypoints, something that an animal or person wouldn't have access to. What I didn't understand for sure is if this information is also available to the robot while performing the "parkour": I'm pretty sure that it perceives the obstacles only through its depth camer…

When the policy is deployed in the real world only the depth camera is used no waypoints etc.. Scan dots and target heading is used in the first Phase of the training to pretrain a policy in simulation. In Phase 2 a policy is trained end-to-end using the pretrained actor network: "First, exteroceptive information is only available in the form of depth images from a front-facing camera instead of scandots. Second, the…

Thank you for your comment. What I don't understand is this: when the robot is in a new environment, how does it know where it's supposed to go? My understanding is that the training teaches the robot how to get to a position, but I didn't see anything about how to choose where to go (in "old AI" parlance, what could have been defined as planning).

Re: Extreme parkour with legged robots

#192

Post by i_am_a_peasant » Wed, Oct 11, 2023, 10:12 AM UTC

ChatGTP wrote: post #133

jupp0r wrote: post #113

Earlier quoted context omitted.

I guess this is why evolution is taking unreasonable amounts of time.

Assuming time has no end, what would be a reasonable amount of time ? Let’s say you could evolve at 10000x the speed ? So what ? Where do you need to be ?

Well at the very least you need to evolve at a rate which can keep up with environmental changes.

Re: Extreme parkour with legged robots

#193

Post by taylodl » Wed, Oct 11, 2023, 1:08 PM UTC

ChatGTP wrote: post #187

taylodl wrote: post #181

Earlier quoted context omitted.

Out of the solar system by the time our Sun dies

7/8 billion years...

Earth's ability to support lifeforms such as human beings is expected to end in the next 500 million to 2 billion years. Granted, that's several orders of magnitude longer than we've been around, but still - the sun will still be shining when humans go extinct, assuming we don't find another home planet and manage to get ourselves to it. It will be billions of years afterward that the sun will actually die.

Re: Extreme parkour with legged robots

#194

Post by ouraf » Wed, Oct 11, 2023, 1:56 PM UTC

one part i didn't fully get is if, after the training, the machine can calculate and implemente the same maneuvers while "offline" (not connected with any other device to assist with the navigation) and what would be the computing hardware requirements for smooth navigation without external help or update? (for example, guiding it with a joystick or GPS waypoint on rugged terrain and having it automatically adjust to each unforeseen obstacle in the way)

Re: Extreme parkour with legged robots

#195

Post by riku_iki » Thu, Oct 12, 2023, 4:52 AM UTC

Cthulhu_ wrote: post #122

riku_iki wrote: post #23

I am always wondering, why they build such complicated biology inspired robots, while they can build something with wheels and simple levels which allows robot to jump over obstacles and turn upside down if it landed on the back.

There are robots / RC cars like that, but they wouldn't work well on rocky surfaces for example; legs are more versatile.

how rocky?.. https://www.youtube.com/watch?v=e6qFzivoUdw