burlap.behavior.singleagent.learning

Interface LearningAgent

All Known Implementing Classes:

ActorCritic, ARTDP, GradientDescentSarsaLam, LSPI, PotentialShapedRMax, QLearning, SarsaLam

public interface LearningAgent

This is the standard interface for defining an agent that learns how to behave in the world through experience. There are two methods that need to be implemented. runLearningEpisode(burlap.mdp.singleagent.environment.Environment) and runLearningEpisode(burlap.mdp.singleagent.environment.Environment, int). Implementing the former method should have the agent interact with the provided Environment until the Environment transitions to a terminal state. The runLearningEpisode(burlap.mdp.singleagent.environment.Environment, int) should have the agent interact with the Environment until either a terminal state is reached or the agent has taken maxSteps in the environment. Both methods should return an Episode object that records the interactions.

Author:

James MacGlashan

Method Summary

Modifier and Type	Method and Description
Episode	runLearningEpisode(Environment env)
Episode	runLearningEpisode(Environment env, int maxSteps)

Method Detail

runLearningEpisode

Episode runLearningEpisode(Environment env)

runLearningEpisode

Episode runLearningEpisode(Environment env,
                           int maxSteps)