And going to skip over most of the history.

Going to try and show you a little bit over the beginnings of planning where the strips

language that we are going to look into, which is still important, has been...

How do I do this so that you can hear anything?

No.

Let's do it this way.

Yes.

I'm not actually interested in trying this out.

So you have to be very quiet.

Otherwise you'll just see.

So this is about the first autonomous robot that planned their own way.

That's in I think 74 or something like that.

45 years ago.

That's probably the most famous robot in the world, Shakey.

Our computers that time, they're about the power of your wristwatch nowadays.

That's what researchers looked like.

Converted hippies.

And that's a printer of that time.

I was having such wonderful interfaces was great at the time.

Shakey, you see?

What is this?

Leaving

GoTo plans and executes a route to a named Go location.

Not to the cross here.

Certain bending obstacles are need be.

The GoTo program plans a route by first computing whether any obstacles lie on a direct path.

The obstacles are set up through tent goals.

After shaking the GoTo, it traverses it to the goal.

Another intermediate level action pushes the named object, box one in this example, to a location specified by their coordinates.

A feature of the intermediate level action is their ability to recover from unforeseen accidents.

Our gremlin, Charlie, symbolizes an agent of change, unknown as shaky.

Shaky sensors tell him that the box has slipped off the push bar.

The cat whiskers tell him where the box is so he can get behind it and resume pushing.

And it goes on and on and on.

So if you want to compare that to the state of the art, go to YouTube and look for RoboCop or something like this.

Where there's robots that look slightly like this, play soccer.

Slightly faster than shaky pushing these blocks around.

So what I want you to understand is that apart from the interesting hardware,

this was the first experiment to do a robot that was kind of autonomous.

Instead of saying, do this, do that, go there, and so on, have these intermediate actions and high level actions.

High level actions we didn't see here, but push the red thing into the room 2468 or something like this.

Or bring a cup of coffee to Mr. So-and-so, which a successor to shaky called Flaky was able to do.

And to compete any of these high level actions, which depend on the context, where is Mr. So-and-so,

and these kind of things you need the planning for.

And you can see in this very complex world, you can't do it with, you cannot do it with atomic states.

You would have to have a state for the block being here, the block being there, the block being there,

shaky being here, shaky being there. This becomes much, much, much too big, becomes even difficult to do names for all these states.

So we need something like planning for that.

Okay, so the tech has gotten better in the last 45 years.

But the problems are essentially the same. The problems are perception.