Now again we can examine the state.

We don't need just something that can change state though. We need something that can also sense the state.

We can convert nondeterministic finite state machines down into deterministic finite state machines.

We can just start at the start state.

Well, how can we get to state 6?

We can say what actions would lead to that state.

And then we can regress this state over this operator and now we have another representation of this state here.

We discussed that problem in terms of saying you're at one state, how can you change that state to get to a neighboring state?

I now can test whether this state is my final goal state.

We see an 'a'. We end up in state 2, and state 2 is an accepting state, so we accept.

In the state space below, shown over here, we can travel between locations

From state 2 on c, we go to state 3.

From state 2 on d, we go to state 3.

And so the state can come in.

And we can denote that by the probability that we get a Result state, S prime given that we start in state, S, and apply action, A.

Remember we can slide around these little tiles and we try to reach a goal state.

If we were on state 2 and we see a 1, we go to state 3.

Now we update this state.

We come in in state 1, and on a we go to state 2.

It can actually find a live state versus a dead state, which is awesome.

And finally, we can also ask the question what happens from a sunny state.

We start here in state 1. We see a 2, so we go over to state 2.

When we have an impasse we can pick a node that looks closer to the goal state.

Can you state the facts to his face?

You can decide, what is your mental state

This can also happen from the resolved state.

Where else can you go in that state?

We can even have a belief state which is an arbitrary formula in Boolean logic, and that can represent anything we want.

Now we can check to see if we can achieve the goal with this two step plan just by checking if there is an intersection between the reachable state and the goal state.

And then every finite state machine can be converted to a deterministic finite state machine.

Suppose we have the parsing state

So we look at the state

So we can check any finite state machine to see if it accepts a string.

What we want to make sure of is that this state can never be reached.

The agent can perceive the state of the environment through its sensors, and it can affect its state through its actuators.

Starting in a start state, on a b we transition to the middle state, on an a we end up in the third state, which is an accepting state.

So we can represent a state, not by an exhaustive listing of everything that's true in the state every single dot, and so on.

Or we decided to add this state based on reasoning about state 2.

Given a particular state we look and see is it a terminal state?

They can only be applied to discrete state spaces.

From every other state that I can think of

From a rainy state we have 2 outbound transitioning, 1 to a sunny state and 1 to a rainy state.

But state 1 is not an accepting state, so we don't accept this string. 'cc3' start here in state 1.

So, remember in problem solving, we start off in a state, and it's a single state, not a belief state.

There's 1 accepting state at state 1, and then we have these few others.