Final Planning Flashcards Preview

CPSC 322: Artificial Intelligence > Final Planning > Flashcards

Flashcards in Final Planning Deck (11):
1

Q1. What is the STRIPS assumption?

Answer: All of the variables not mentioned in the description of an action stay unchanged when the action is carried out.

2

Q2. What are the two parts of an action in STRIPS, and what is the meaning of each part?

Answer: Preconditions - a set of assignments of values to variables that must be true for the action to occur. Effects - a set of resulting assignments of values to those variables that change as the result of the action.

3

Q3. What is a frame rule?

Answer: It is a rule in the representation of a planning problem which specifies the conditions under which a variable maintains its value from one time step to the next.

4

Q4. In forward planning, what is the branching factor for a given state?

Answer: In forward planning, the branching factor for a given state is the number of actions possible at this state (those that have their preconditions satisfied).

5

Q5. Explain the role of the horizon in CSP planning. How does the choice of horizon affect the behavior of the planner?

Answer: In order to represent a plan as a CSP, we need to commit to a maximal length for the plan. This length is called the horizon. If the horizon is too short, the planner will not be able to find a plan, even though the planning problem is not impossible. Having a larger horizon than necessary will make the planning problem take much longer to solve.

6

Q6. In a CSP encoding of a planning problem, what is the relationship between the horizon (h), the number of variables in the CSP (n), and the number of variables (v) and actions (a) in the planning problem?

Answer: n = (h+1)*v + a*h

7

Q7. In a CSP encoding of a planning problem, which variables are involved in an effect constraint?

Answer: State variables at time t, actions at time t and state variables at time t + 1.

8

Q8. In a CSP encoding of a planning problem, which variables are involved in a precondition constraint?

Answer: State variables at time t and actions at time t.

9

Q9. In a CSP encoding of a planning problem, why does an effect constraint involve both state variables at the current time step and state variables at the next time step? (That is, why is it not enough simply to constrain the values of the variables at the next time step whose values are changed by the given action?)

Answer: ?

10

Q10. Describe one way of mapping a planning problem into a graph search problem. What is a node? What is the successor function? What is a start node? What is the goal test?

Answer:? A node is an assignment of CSP variables, and successor function are actions.

11

QA11. What are some key limitations of STRIPS?

Answer: States are represented simply as a conjunction of positive literals, e.g. poor ∧ unknown, goals are conjunctions (no disjunction allowed), no support for equality.