All Downloads are FREE. Search and download functionalities are using the official Maven repository.
  • Log In
  • Register

    • burlap.behavior.singleagent.interfaces.rlglue.RLGlueState Maven / Gradle / Ivy

    Go to download

    The Brown-UMBC Reinforcement Learning and Planning (BURLAP) Java code library is for the use and development of single or multi-agent planning and learning algorithms and domains to accompany them. The library uses a highly flexible state/observation representation where you define states with your own Java classes, enabling support for domains that discrete, continuous, relational, or anything else. Planning and learning algorithms range from classic forward search planning to value-function-based stochastic planning and learning algorithms.

    The newest version!
    package burlap.behavior.singleagent.interfaces.rlglue;

import burlap.mdp.core.state.State;
import org.rlcommunity.rlglue.codec.types.Observation;

import java.util.ArrayList;
import java.util.List;

/**
 * A {@link State} for RLGLue {@link Observation} objects. Each instance contains the inner RLGlue {@link Observation}
 * You can set and get the {@link Observation} with standard methods, thereby allowing serialization. The BURLAP
 * variable keys are specified with the {@link RLGlueVarKey}, which indicates indices for character, integer, and double
 * values that RLGlue supports. You can also get variable values using String keys of the form "cn", "in", "dn" for character
 * integer, and double variables respectively, where n is the index into that vector.
 * @author James MacGlashan.
 */
public class RLGlueState implements State{

	protected Observation obs;

	public RLGlueState() {
	}

	public RLGlueState(Observation obs) {
		this.obs = obs;
	}

	@Override
	public List variableKeys() {
		List keys = new ArrayList(obs.getNumChars() + obs.getNumInts() + obs.getNumDoubles());
		for(int i = 0; i < obs.getNumChars(); i++){
			RLGlueVarKey key = new RLGlueVarKey('c', i);
			keys.add(key);
		}
		for(int i = 0; i < obs.getNumInts(); i++){
			RLGlueVarKey key = new RLGlueVarKey('i', i);
			keys.add(key);
		}
		for(int i = 0; i < obs.getNumDoubles(); i++){
			RLGlueVarKey key = new RLGlueVarKey('d', i);
			keys.add(key);
		}
		return keys;
	}

	@Override
	public Object get(Object variableKey) {
		RLGlueVarKey key;
		if(variableKey instanceof RLGlueVarKey){
			key = (RLGlueVarKey)variableKey;
		}
		else if(variableKey instanceof String){
			key = new RLGlueVarKey((String)variableKey);
		}
		else{
			throw new RuntimeException("RLGlueState does not understand key of type " + variableKey.getClass().getName() + "; use a RLGlueVarKey or String rep");
		}

		if(key.type == 'c'){
			return obs.charArray[key.ind];
		}
		else if(key.type == 'i'){
			return obs.intArray[key.ind];
		}
		else if(key.type == 'd'){
			return obs.doubleArray[key.ind];
		}

		return null;
	}

	@Override
	public State copy() {
		return new RLGlueState(obs); //immutable so we can do shallow copy
	}

	public Observation getObs() {
		return obs;
	}

	public void setObs(Observation obs) {
		this.obs = obs;
	}

	public static class RLGlueVarKey{
		public char type;
		public int ind;


		public RLGlueVarKey(char type, int ind) {
			this.type = type;
			this.ind = ind;
		}

		public RLGlueVarKey(String strRep){
			this.type = strRep.charAt(0);
			this.ind = Integer.parseInt(strRep.substring(1));
		}

		@Override
		public String toString() {
			return String.valueOf(type) + ind;
		}

		@Override
		public boolean equals(Object o) {
			if(this == o) return true;
			if(o == null || getClass() != o.getClass()) return false;

			RLGlueVarKey that = (RLGlueVarKey) o;

			if(type != that.type) return false;
			return ind == that.ind;

		}

		@Override
		public int hashCode() {
			int result = (int) type;
			result = 31 * result + ind;
			return result;
		}
	}
}
    
    
    
    

    




    
    
    © 2015 - 2024 Weber Informatics LLC | Privacy Policy