edu.columbia.cs.coms4701.search.online
Interface OnlineSearchProblemWithSuccessorFunction<K extends AgentAction,S extends AgentState>

All Superinterfaces:
OnlineSearchProblem<K,S>

public interface OnlineSearchProblemWithSuccessorFunction<K extends AgentAction,S extends AgentState>
extends OnlineSearchProblem<K,S>

An online problem description in which the agent does know, at least provisionally, the outcome of a given action from a given state.

An agent may formulate an OnlineSearchProblemWithSuccessorFunction if it knows something about how the environment should work; for example, if it knows that (most of the time) going North from a particular position will result in a state in which its position has moved up 1 square.

Note that in an online environment, the successor function may be "wrong"; that is, it may return (action, succesor) pairs that, when the action is performed, turn out not to be the case. If this happens, the agent should update the successor function using the updateSuccessorFunction method.

Version:
0.1

User: aaron Date: Sep 12, 2006 Time: 8:48:57 AM

Author:
aaron@cs

Method Summary
 java.util.Map<K,S> successorFunction(S theState)
          Given a particular state x, returns a set of (action, successor) ordered pairs, where each action is one of the legal actions in state x and each successor is a state that can be reached from x by applying the action.
 void updateSuccessorFunction(S firstState, K theAction, S secondState)
          Accessor to update the problem's successor function with new information learned from the environment.
 
Methods inherited from interface edu.columbia.cs.coms4701.search.online.OnlineSearchProblem
actions, goalTest, stepCost
 

Method Detail

successorFunction

java.util.Map<K,S> successorFunction(S theState)
Given a particular state x, returns a set of (action, successor) ordered pairs, where each action is one of the legal actions in state x and each successor is a state that can be reached from x by applying the action.

Note that in an online environment, the successor function may be "wrong"; that is, it may return (action, succesor) pairs that, when the action is performed, turn out not to be the case. If this happens, the agent should update the successor function using the method below.

Parameters:
theState -
Returns:
set of possible actions and their resulting states

updateSuccessorFunction

void updateSuccessorFunction(S firstState,
                             K theAction,
                             S secondState)
Accessor to update the problem's successor function with new information learned from the environment.

The necessity of updating the problem description after actions have been performed and percepts received distinguishes an online problem from an ordinary search problem.

Parameters:
firstState -
theAction -
secondState -