本文整理汇总了Java中burlap.mdp.core.action.UniversalActionType类的典型用法代码示例。如果您正苦于以下问题:Java UniversalActionType类的具体用法?Java UniversalActionType怎么用?Java UniversalActionType使用的例子?那么恭喜您, 这里精选的类代码示例或许可以为您提供帮助。
UniversalActionType类属于burlap.mdp.core.action包,在下文中一共展示了UniversalActionType类的14个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于我们的系统推荐出更棒的Java代码示例。
示例1: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public SADomain generateDomain() {
SADomain domain = new SADomain();
domain.addActionTypes(
new UniversalActionType(ACTION_NORTH),
new UniversalActionType(ACTION_SOUTH),
new UniversalActionType(ACTION_EAST),
new UniversalActionType(ACTION_WEST));
GridWorldStateModel smodel = new GridWorldStateModel();
RewardFunction rf = new ExampleRF();
TerminalFunction tf = new ExampleTF();
domain.setModel(new FactoredModel(smodel, rf, tf));
return domain;
}
开发者ID:honzaMaly,项目名称:kusanagi,代码行数:21,代码来源:ExampleGridWorld.java
示例2: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public SADomain generateDomain() {
SADomain domain = new SADomain();
domain.addActionTypes(
new UniversalActionType(NextActionEnumerations.YES.name()),
new UniversalActionType(NextActionEnumerations.NO.name()));
//unknown reward
RewardFunction rf = (state, action, state1) -> defaultReward;
//no terminal state
TerminalFunction tf = state -> false;
domain.setModel(new FactoredModel(model, rf, tf));
return domain;
}
开发者ID:honzaMaly,项目名称:kusanagi,代码行数:18,代码来源:DecisionDomainGenerator.java
示例3: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public SADomain generateDomain() {
SADomain domain = new SADomain();
domain.addActionTypes(
new UniversalActionType(ACTION_NORTH),
new UniversalActionType(ACTION_SOUTH),
new UniversalActionType(ACTION_EAST),
new UniversalActionType(ACTION_WEST));
GridWorldStateModel smodel = new GridWorldStateModel();
RewardFunction rf = new ExampleRF(this.goalx, this.goaly);
TerminalFunction tf = new ExampleTF(this.goalx, this.goaly);
domain.setModel(new FactoredModel(smodel, rf, tf));
return domain;
}
开发者ID:jmacglashan,项目名称:burlap_examples,代码行数:21,代码来源:ExampleGridWorld.java
示例4: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public SADomain generateDomain() {
SADomain domain = new SADomain();
// add in NullActions for Domain
for (String actionName : actionNames) {
domain.addActionType(new UniversalActionType(new ALEAction(actionName)));
}
return domain;
}
开发者ID:h2r,项目名称:burlap_ale,代码行数:12,代码来源:ALEDomainGenerator.java
示例5: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public Domain generateDomain() {
SGDomain domain = new SGDomain();
for(int i = 0; i < this.actionNameToIndex.length; i++){
for(Map.Entry<String, Integer> as : this.actionNameToIndex[i].namesToInd.entrySet()){
domain.addActionType(new UniversalActionType(i + "_" + as.getKey(), new MatrixAction(as.getKey(), as.getValue())));
}
}
domain.setJointActionModel(new StaticRepeatedGameModel());
return domain;
}
开发者ID:jmacglashan,项目名称:burlap,代码行数:16,代码来源:SingleStageNormalFormGame.java
示例6: generateAgentType
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
public SGAgentType generateAgentType(int player){
List<ActionType> actions = new ArrayList<ActionType>();
for(Map.Entry<String, Integer> e : actionNameToIndex[player].namesToInd.entrySet()){
actions.add(new UniversalActionType(player + e.getKey(), new MatrixAction(e.getKey(), e.getValue())));
}
SGAgentType type = new SGAgentType("player" + player, actions);
return type;
}
开发者ID:jmacglashan,项目名称:burlap,代码行数:9,代码来源:SingleStageNormalFormGame.java
示例7: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public SADomain generateDomain() {
SADomain domain = new SADomain();
IPPhysicsParams cphys = this.physParams.copy();
IPModel smodel = new IPModel(cphys);
RewardFunction rf = this.rf;
TerminalFunction tf = this.tf;
if(rf == null){
rf = new InvertedPendulumRewardFunction();
}
if(tf == null){
tf = new InvertedPendulumTerminalFunction();
}
FactoredModel model = new FactoredModel(smodel, rf ,tf);
domain.setModel(model);
domain.addActionType(new UniversalActionType(ACTION_LEFT))
.addActionType(new UniversalActionType(ACTION_RIGHT))
.addActionType(new UniversalActionType(ACTION_NO_FORCE));
return domain;
}
开发者ID:jmacglashan,项目名称:burlap,代码行数:30,代码来源:InvertedPendulum.java
示例8: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public OOSADomain generateDomain() {
OOSADomain domain = new OOSADomain();
domain.addStateClass(CLASS_AGENT, ExGridAgent.class)
.addStateClass(CLASS_LOCATION, EXGridLocation.class);
domain.addActionTypes(
new UniversalActionType(ACTION_NORTH),
new UniversalActionType(ACTION_SOUTH),
new UniversalActionType(ACTION_EAST),
new UniversalActionType(ACTION_WEST));
OODomain.Helper.addPfsToDomain(domain, this.generatePfs());
OOGridWorldStateModel smodel = new OOGridWorldStateModel();
RewardFunction rf = new SingleGoalPFRF(domain.propFunction(PF_AT), 100, -1);
TerminalFunction tf = new SinglePFTF(domain.propFunction(PF_AT));
domain.setModel(new FactoredModel(smodel, rf, tf));
return domain;
}
开发者ID:jmacglashan,项目名称:burlap_examples,代码行数:27,代码来源:ExampleOOGridWorld.java
示例9: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public OOSGDomain generateDomain() {
OOSGDomain domain = new OOSGDomain();
domain.addStateClass(CLASS_AGENT, GGAgent.class)
.addStateClass(CLASS_GOAL, GGGoal.class)
.addStateClass(CLASS_DIM_H_WALL, GGWall.GGHorizontalWall.class)
.addStateClass(CLASS_DIM_V_WALL, GGWall.GGVerticalWall.class);
domain.addActionType(new UniversalActionType(ACTION_NORTH))
.addActionType(new UniversalActionType(ACTION_SOUTH))
.addActionType(new UniversalActionType(ACTION_EAST))
.addActionType(new UniversalActionType(ACTION_WEST))
.addActionType(new UniversalActionType(ACTION_NOOP));
OODomain.Helper.addPfsToDomain(domain, this.generatePFs());
domain.setJointActionModel(new GridGameStandardMechanics(domain, this.semiWallProb));
return domain;
}
开发者ID:jmacglashan,项目名称:burlap,代码行数:27,代码来源:GridGame.java
示例10: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public SADomain generateDomain() {
SADomain domain = new SADomain();
MCModel smodel = new MCModel(this.physParams.copy());
if(tf == null){
tf = new ClassicMCTF(physParams.xmax);
}
if(rf == null){
rf = new GoalBasedRF(tf, 100, 0);
}
FactoredModel model = new FactoredModel(smodel, rf, tf);
domain.setModel(model);
domain.addActionType(new UniversalActionType(ACTION_FORWARD))
.addActionType(new UniversalActionType(ACTION_BACKWARDS))
.addActionType(new UniversalActionType(ACTION_COAST));
return domain;
}
开发者ID:jmacglashan,项目名称:burlap,代码行数:27,代码来源:MountainCar.java
示例11: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public OOSADomain generateDomain() {
OOSADomain domain = new OOSADomain();
List <Double> thrustValuesTemp = this.thrustValues;
if(thrustValuesTemp.isEmpty()){
thrustValuesTemp.add(0.32);
thrustValuesTemp.add(-physParams.gravity);
}
domain.addStateClass(CLASS_AGENT, LLAgent.class)
.addStateClass(CLASS_PAD, LLBlock.LLPad.class)
.addStateClass(CLASS_OBSTACLE, LLBlock.LLObstacle.class);
//make copy of physics parameters
LLPhysicsParams cphys = this.physParams.copy();
//add actions
domain.addActionType(new UniversalActionType(ACTION_TURN_LEFT))
.addActionType(new UniversalActionType(ACTION_TURN_RIGHT))
.addActionType(new UniversalActionType(ACTION_IDLE))
.addActionType(new ThrustType(thrustValues));
OODomain.Helper.addPfsToDomain(domain, this.generatePfs());
LunarLanderModel smodel = new LunarLanderModel(cphys);
RewardFunction rf = this.rf;
TerminalFunction tf = this.tf;
if(rf == null){
rf = new LunarLanderRF(domain);
}
if(tf == null){
tf = new LunarLanderTF(domain);
}
FactoredModel model = new FactoredModel(smodel, rf, tf);
domain.setModel(model);
return domain;
}
开发者ID:jmacglashan,项目名称:burlap,代码行数:44,代码来源:LunarLanderDomain.java
示例12: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
/**
* Creates a new frostbite domain.
*
* @return the generated domain object
*/
@Override
public OOSADomain generateDomain() {
OOSADomain domain = new OOSADomain();
domain.addStateClass(CLASS_AGENT, FrostbiteAgent.class)
.addStateClass(CLASS_IGLOO, FrostbiteIgloo.class)
.addStateClass(CLASS_PLATFORM, FrostbitePlatform.class);
//add actions
domain.addActionType(new UniversalActionType(ACTION_NORTH))
.addActionType(new UniversalActionType(ACTION_SOUTH))
.addActionType(new UniversalActionType(ACTION_EAST))
.addActionType(new UniversalActionType(ACTION_WEST))
.addActionType(new UniversalActionType(ACTION_IDLE));
//add pfs
List<PropositionalFunction> pfs = this.generatePFs();
for(PropositionalFunction pf : pfs){
domain.addPropFunction(pf);
}
FrostbiteModel smodel = new FrostbiteModel(scale);
RewardFunction rf = this.rf;
TerminalFunction tf = this.tf;
if(rf == null){
rf = new FrostbiteRF(domain);
}
if(tf == null){
tf = new FrostbiteTF(domain);
}
FactoredModel model = new FactoredModel(smodel, rf, tf);
domain.setModel(model);
return domain;
}
开发者ID:jmacglashan,项目名称:burlap,代码行数:47,代码来源:FrostbiteDomain.java
示例13: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public SADomain generateDomain() {
SADomain domain = new SADomain();
CPPhysicsParams cphys = this.physParams.copy();
RewardFunction rf = this.rf;
TerminalFunction tf = this.tf;
if(rf == null){
rf = new CartPoleRewardFunction();
}
if(tf == null){
tf = new CartPoleTerminalFunction();
}
FullStateModel smodel = cphys.useCorrectModel ? new CPClassicModel(cphys) : new CPClassicModel(cphys);
FactoredModel model = new FactoredModel(smodel, rf, tf);
domain.setModel(model);
domain.addActionType(new UniversalActionType(ACTION_LEFT))
.addActionType(new UniversalActionType(ACTION_RIGHT));
return domain;
}
开发者ID:jmacglashan,项目名称:burlap,代码行数:30,代码来源:CartPoleDomain.java
示例14: generateDomain
import burlap.mdp.core.action.UniversalActionType; //导入依赖的package包/类
@Override
public Domain generateDomain() {
PODomain domain = new PODomain();
domain.addActionType(new UniversalActionType(ACTION_LEFT))
.addActionType(new UniversalActionType(ACTION_RIGHT))
.addActionType(new UniversalActionType(ACTION_LISTEN));
if(this.includeDoNothing){
domain.addActionType(new UniversalActionType(ACTION_DO_NOTHING));
}
ObservationFunction of = new TigerObservations(this.listenAccuracy, this.includeDoNothing);
domain.setObservationFunction(of);
TigerModel model = new TigerModel(correctDoorReward, wrongDoorReward, listenReward, nothingReward);
domain.setModel(model);
StateEnumerator senum = new StateEnumerator(domain, new SimpleHashableStateFactory());
senum.getEnumeratedID(new TigerState(VAL_LEFT));
senum.getEnumeratedID(new TigerState(VAL_RIGHT));
domain.setStateEnumerator(senum);
return domain;
}
开发者ID:jmacglashan,项目名称:burlap,代码行数:31,代码来源:TigerDomain.java
注:本文中的burlap.mdp.core.action.UniversalActionType类示例整理自Github/MSDocs等源码及文档管理平台,相关代码片段筛选自各路编程大神贡献的开源项目,源码版权归原作者所有,传播和使用请参考对应项目的License;未经允许,请勿转载。 |
请发表评论