1. The document discusses algorithms for planning as state-space search, including forward and backward search.
2. Forward search is prone to exploring irrelevant actions, making even small planning problems intractable without an accurate heuristic.
3. Backward search starts from the goal state and applies actions in reverse to find a sequence reaching the initial state, avoiding exploring irrelevant actions. It uses regression to determine predecessor states from goals and actions.
1. The document discusses algorithms for planning as state-space search, including forward and backward search.
2. Forward search is prone to exploring irrelevant actions, making even small planning problems intractable without an accurate heuristic.
3. Backward search starts from the goal state and applies actions in reverse to find a sequence reaching the initial state, avoiding exploring irrelevant actions. It uses regression to determine predecessor states from goals and actions.
1. The document discusses algorithms for planning as state-space search, including forward and backward search.
2. Forward search is prone to exploring irrelevant actions, making even small planning problems intractable without an accurate heuristic.
3. Backward search starts from the goal state and applies actions in reverse to find a sequence reaching the initial state, avoiding exploring irrelevant actions. It uses regression to determine predecessor states from goals and actions.
PLANNING AS STATE-SPACE SEARCH Forward state space search(Progression)
Consider the task of buying a book with 10 ISBN
numbers. The ISBN 0f AI has 10 digits-so there are 10 to the power 10 actions from initial state(Uninformed search). The planning problems have large state space. Forward search is prone to exploring irrelevant actions. Without an accurate heuristic even a small planning problem is hopeless in forward search. Backward(regression)relevant states search • We start with the goal and apply the actions backward until we find a sequence of steps that reaches the initial state. Start with goal-conjunction of literals For example my goal is a conjunction of fluents like ¬poor(john)^Famous(John),then it describes the goal state where poor is false and famous is true. Suppose if there are n ground fluents in a domain,thene there are 2^n ground states,3^n descriptions of set of goal states.(true,false or not mentioned). Backward regression contd.. • The PDDL was designed to move backwards easily • a ground goal description g and a ground action a, the regression from g over a gives us a state description g’ defined by • g’ =( g − ADD(a))∪Precond(a) . • the effects that were added by the actions were not true in previous state and the preconditions for the goal were true in the previous state. • The goal is to deliver a specific piece of cargo to SFO: • consider the goal Own(0136042597), • given an initial state with 10 billion ISBNs, • and the single action A = Action(Buy(i),PRECOND:ISBN(i),EFFECT:Own(i)).
• forward search without a heuristic would have to start enumerating
the 10 billion ground Buy actions. • But with backward search, we would unify the goal Own(0136042597) with the (standardized) effect Own(x), yielding the substitution θ = {x/0136042597}. • Then we would regress over the action Subst(θ,A’) to yield the predecessor state description ISBN(0136042597) Graph planning-mutex links What is a Frame Problem • Environment for an agent is not static • Identifying which things remain static in changing word • Use of information in Knowledge Base and agent’s conclusion based on that knowledge • Discarding negative inferences