burlap.behavior.singleagent.planning.deterministic.informed

Interface Heuristic

All Known Implementing Classes:

NullHeuristic

public interface Heuristic

An interface for defining heuristics. The heuristic function should return an estimate of the amount of *reward* that will be accumulated from that given state. Since deterministic forward search planning algorithms typically expect costs, this is represented by simply using negative reward, where values closer to zero are better. For instance, if it was known that a state was 3 steps away from the goal, an optimal heuristic (and the true cost from the state) would return -3.

Author:

James MacGlashan

Method Summary

Modifier and Type	Method and Description
double	h(State s) Returns the estimated amount of reward that will be received when following the optimal policy from the given state.

Method Detail

h

double h(State s)

Returns the estimated amount of reward that will be received when following the optimal policy from the given state. Since deterministic forward search planning algorithms typically expect costs, this is represented by simply using negative reward, where values closer to zero are better. For instance, if it was known that state s was 3 steps away from the goal, an optimal heuristic (the true reward from the state) would return -3.

Parameters:

s - the state from which to estimate the future reward.

Returns:

the estimated amount of reward that will be received when following the optimal policy from s.