I'm going to talk about strips.

Okay.

Shecky was made at the SRI International Stanford Research Institute.

It's kind of an independent think tank next to Stanford.

So that's the SRI Problem Solver strips.

And essentially it uses propositional logic to describe the world.

Which of course is reasonable.

You remember that the space in which Shecky lived was kind of a suite of rooms.

That kind of looked like this, had a couple of doors, probably something like this.

And so you kind of had a finite number of rooms, just like one person.

And then you had a couple of blocks that were in the rooms and maybe in some quadrants and so on.

So you could number all locations.

You knew about all of the blocks. They all had a name and so on.

So you could say something like at Shecky X32, where X32 or L32.

And that was something you actually saw in the print house where Shecky could be.

So this here is really a propositional variable.

So it's fine to do that. So we use propositional logic.

We need, as you saw, a way of talking about preconditions, effects and goals.

Goals to know whether we're in a goal state.

Preconditions to know whether an action applies and effects of what an action achieves.

And the preconditions and goals we basically write down as conjunctions of positive atoms.

It turns out that in this thing that is enough.

In many places by just choosing wisely what your propositional variables are,

whether you have a propositional variable for free, something is like in the blocks world, something is free on top.

Or you say it is not the case as though something is on there.

If you choose your vocabulary wisely, then you can actually get by with conjunctions of positive atoms, which makes certain things easier.

And then the effects are usually, you use literals for that because you need updates.

I'm in room X42 now and I'm no longer, take the negatives for that, in room something or the other.

So that's really what's behind this. We have Boolean variables, we have preconditions, and we have effects which are positive and negative.

And that is essentially paired with a special notation.

So that was invented for shaky.

So we have a strips planning task which looks pretty similar to what we had as problems all the way in the beginning of the course.

Only that instead of talking about states and successors and all of those kind of things, we describe the world by a finite set of facts.

Block so-and-so is here, shaky is in the room so-and-so. You should move the red block to something or the other.

Then we have finite sets of actions.

Each action is characterized by three lists of facts.

The first one is the precondition list, which is what are the things that must be true so that I can actually make an action.

And then we have an add and a delete list. Add is a set of facts we add to the current world.

And delete is a set of things we delete from the world description. Those are the down dates if you want.

And that is something that is very, very important.

Because if you think about it, you could also describe actions just by describing them.

But then you would have a tremendous problem because we have what we call the frame problem.

Every action really only has very limited effects.

If I close my computer here, you won't see anything anymore and that would disrupt the video recording of the slides.

But it will not actually change American politics, probably.

So for every action you have a certain effect. Video goes to the fritz and you don't see anything anymore.

But even more things that are unchanged.

Almost the gazillion of things, of facts, that are unchanged. Also, my closing computer doesn't change the laws of arithmetic or physics.

Almost everything is unchanged and we rely on that.

Now, you can try and write that down in logic and you'll see that this becomes very, very difficult.