Class AbstractAgent

java.lang.Object
  |
  +--java.lang.Thread
        |
        +--AbstractAgent

All Implemented Interfaces:

Runnable

Direct Known Subclasses:

PlanningAgent

public abstract class AbstractAgent

extends Thread

This class provides the basic structure for the implementation of agent loops. All implementations of agent loops must be (possibly indirect) subclasses of this class.

Agents can be customized by users of the APLAgent Manager by modifying the agent's parameters via the GUI. The set of parameters of an agent is described by the appropriate subclass of AgentParams. When subclasses of AbstractAgent invoke the constructor of AbstractAgent, they may specify the name of the parameter class that they intend to use. For example:

    private ExampleAgentParams myParams;

    public ExampleAgent() throws Exception
    {     super("ExampleAgentParams");

          // APLAgent Manager creates an instance for us
          myParams=(ExampleAgentParams)getAgentParams();
    }
 

The agent loop is implemented by method agentLoop(). Subclasses of AbstractAgent are expected to define their agent loop. For example:

	 // The agent loop for the planning agent.
	public void agentLoop() throws Exception
	{	ActionSequence plan;
		String goal;

		goal=getObservationsAndGoal();

		// if agent sets goal autonomously, use: getObservations() and showSelectedGoal(goal)

		plan=findPlan(goal);

		performPartOfPlan(plan);

	}
 

Calls to the inference engine should be performed by means of methods computeOneTrajectory() and computeAllTrajectory(). The list of files that are required by these methods can be retrieved by applying method getFiles on fields aFiles (agent files) and dFiles (domain files). For example, a code fragment that computes one plan is:

			MathSet files=new MathSet();
			files.add(tempFile);
			files=files.union(dFiles.getFiles(DomainFiles.ACT_DESC));
			files=files.union(dFiles.getFiles(DomainFiles.PROB_INST));
			files=files.union(dFiles.getFiles(DomainFiles.HIDES));
			files=files.union(aFiles.getFiles(AgentFiles.PLANNING));
			files=files.union(aFiles.getFiles(AgentFiles.HIDES));
			trj=computeOneTrajectory(files,currTime,lastTime);
 

Field Summary

MathSet	actions The set of actions defined in the domain.
AgentFiles	aFiles This is the collection of A-Prolog files for the reasoning modules of the agent.
MathSet	agentActions The set of agent actions defined in the domain.
int	currTime The current time step.
DomainFiles	dFiles This is the collection of A-Prolog files that describe the domain.
Engine	engine Used internally
MathSet	exogenousActions The set of exogenous actions defined in the domain.
MathSet	fluents The set of fluents defined in the domain.
MathSet	goals The set of goals defined in the domain.
History	history The current history.
String	tempFile The name of the temporary file.

Fields inherited from class java.lang.Thread

MAX_PRIORITY, MIN_PRIORITY, NORM_PRIORITY

Constructor Summary

AbstractAgent()
Constructs an AbstractAgent, with the default set of parameters.

AbstractAgent(String agentParamsClassName)
Constructs an AbstractAgent, using the parameter class whose name is given.

Method Summary

abstract void	agentLoop() The agent loop.
void	appendHistory(String tf) Appends the current history to the temporary file.
void	appendPlan(String tf, ActionSequence plan, int currTime) Appends a plan (i.e., an ActionSequence) to the temporary file, starting from the given time step.
void	appendText(String tf, String lit) Appends a string to the temporary file, returned by a newline character.
void	clearTempFile(String tf) Deletes the contents of the given temporary file.
TrajectorySet	computeAllTrajectories(MathSet fileList, int currTime, int lastTime) Computes all the models of the given list of files, with the given values for A-Prolog parameters currtime and lasttime, and returns the TrajectorySet represented by the models.
Trajectory	computeOneTrajectory(MathSet fileList, int currTime, int lastTime) Computes one model of the given list of files, with the given values for A-Prolog parameters currtime and lasttime, and returns the Trajectory contained in the model.
TrajectorySet	findDefinedTrajectories(ActionSequence plan) Computes the set of trajectories defined by the given plan.
AgentParams	getAgentParams() Returns the set of parameters for the agent, after they have been initialized by the APLAgent Manager.
void	getObservations() Gathers the observations about the current time step.
void	getObservations(boolean goalSelection) Gathers the observations about the current time step.
String	getObservationsAndGoal() Gathers the observations about the current time step.
void	initialize(APLAgentMgr p, AgentFiles af, DomainFiles df, History h, Vector g, File tf, Engine e, AgentParams ap) Invoked only by the APLAgent Manager
void	log(String s) Appends a line to the Log Window in the APLAgent Manager.
void	performPartOfPlan(ActionSequence plan) Performs the first compound action of the given plan.
void	repeatIteration() Repeats the last iteration of the agent loop.
void	repeatIterationNoModel() Repeats the last iteration of the agent loop.
void	repeatIterationNoPlan() Repeats the last iteration of the agent loop.
void	run() Invoked only by the APLAgent Manager
void	showSelectedGoal(String goal) Updates the GUI in order to show the goal selected by the agent.
void	terminateLoop() Terminates the agent loop.
void	terminateLoopNoModel() Terminates the agent loop.
void	terminateLoopNoPlan() Terminates the agent loop.
void	updateAgentFiles(AgentFiles af) Invoked only by the APLAgent Manager
void	updateGoals(Vector g, DomainFiles df) Invoked only by the APLAgent Manager

Methods inherited from class java.lang.Thread

activeCount, checkAccess, countStackFrames, currentThread, destroy, dumpStack, enumerate, getContextClassLoader, getName, getPriority, getThreadGroup, interrupt, interrupted, isAlive, isDaemon, isInterrupted, join, join, join, resume, setContextClassLoader, setDaemon, setName, setPriority, sleep, sleep, start, stop, stop, suspend, toString, yield

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

aFiles

public AgentFiles aFiles

This is the collection of A-Prolog files for the reasoning modules of the agent.

dFiles

public DomainFiles dFiles

This is the collection of A-Prolog files that describe the domain.

engine

public Engine engine

Used internally

history

public History history

The current history.

fluents

public MathSet fluents

The set of fluents defined in the domain.

actions

public MathSet actions

The set of actions defined in the domain. (It is the union of agent actions and exogenous actions.)

agentActions

public MathSet agentActions

The set of agent actions defined in the domain.

exogenousActions

public MathSet exogenousActions

The set of exogenous actions defined in the domain.

goals

public MathSet goals

The set of goals defined in the domain.

currTime

public int currTime

The current time step.

tempFile

public String tempFile

The name of the temporary file. This file can be used by the agent to prepare data that must be fed to the inference engine.

Constructor Detail

AbstractAgent

public AbstractAgent()

Constructs an AbstractAgent, with the default set of parameters.

AbstractAgent

public AbstractAgent(String agentParamsClassName)
              throws ClassNotFoundException,
                     InstantiationException,
                     IllegalAccessException

Constructs an AbstractAgent, using the parameter class whose name is given.

Parameters:

agentParamsClassName - the name of the parameter class

Method Detail

initialize

public void initialize(APLAgentMgr p,
                       AgentFiles af,
                       DomainFiles df,
                       History h,
                       Vector g,
                       File tf,
                       Engine e,
                       AgentParams ap)

Invoked only by the APLAgent Manager

updateAgentFiles

public void updateAgentFiles(AgentFiles af)

Invoked only by the APLAgent Manager

updateGoals

public void updateGoals(Vector g,
                        DomainFiles df)

Invoked only by the APLAgent Manager

getAgentParams

public AgentParams getAgentParams()

Returns the set of parameters for the agent, after they have been initialized by the APLAgent Manager.

Returns:

the set of parameters for the agent

clearTempFile

public void clearTempFile(String tf)

Deletes the contents of the given temporary file.

Parameters:

tf - the name of the temporary file

appendHistory

public void appendHistory(String tf)

Appends the current history to the temporary file.

Parameters:

tf - the name of the temporary file

appendPlan

public void appendPlan(String tf,
                       ActionSequence plan,
                       int currTime)

Appends a plan (i.e., an ActionSequence) to the temporary file, starting from the given time step.

Parameters:

tf - the name of the temporary file

plan - the plan to be appended to the file

currTime - the first time step to be appended

appendText

public void appendText(String tf,
                       String lit)

Appends a string to the temporary file, returned by a newline character.

Parameters:

tf - the name of the temporary file

lit - the string to be appended

log

public void log(String s)

Appends a line to the Log Window in the APLAgent Manager.

Parameters:

s - the line to be appended

getObservations

public void getObservations(boolean goalSelection)
                     throws Exception

Gathers the observations about the current time step. This is the method that agent loops are expected to invoke for the observation-gathering step. Goal selection by the user can be either enabled or disabled, depending on whether the agent is able to pick goals autonomously.

When the method returns, observations can be found in the history field of the AbstractAgent class.

Parameters:

goalSelection - true if goal selection is allowed; false otherwise

getObservations

public void getObservations()
                     throws Exception

Gathers the observations about the current time step. This is the method that agent loops are expected to invoke for the observation-gathering step. Goal selection by the user is not allowed, which means that the agent is expected to be able to pick goals autonomously.

When the method returns, observations can be found in the history field of the AbstractAgent class.

getObservationsAndGoal

public String getObservationsAndGoal()
                              throws Exception

Gathers the observations about the current time step. This is the method that agent loops are expected to invoke for the observation-gathering step. The user is also expected to select the agent's goal, which is returned by the method. When the method returns, observations can be found in the history field of the AbstractAgent class.

Returns:

the agent's goal

showSelectedGoal

public void showSelectedGoal(String goal)

Updates the GUI in order to show the goal selected by the agent.

Parameters:

goal - the goal selected by the agent

performPartOfPlan

public void performPartOfPlan(ActionSequence plan)
                       throws Exception

Performs the first compound action of the given plan. This is the method that agent loops are expected to invoke for the plan-execution step.

Parameters:

plan - the plan whose first compoud action must be performed

findDefinedTrajectories

public TrajectorySet findDefinedTrajectories(ActionSequence plan)
                                      throws Exception

Computes the set of trajectories defined by the given plan.

Parameters:

plan - the plan that defines the trajectories

Returns:

a TrajectorySet containing the trajectories defined by the plan

terminateLoop

public void terminateLoop()
                   throws TerminateLoopException

Terminates the agent loop. This method will cause the agent to abort the current step of the agent loop and halt execution.

terminateLoopNoPlan

public void terminateLoopNoPlan()
                         throws Exception

Terminates the agent loop. This method will cause the agent to abort the current step of the agent loop and halt execution. A dialog will be displayed, saying that no plans were returned by the inference engine.

terminateLoopNoModel

public void terminateLoopNoModel()
                          throws Exception

Terminates the agent loop. This method will cause the agent to abort the current step of the agent loop and halt execution. A dialog will be displayed, saying that no models were returned by the inference engine.

repeatIterationNoPlan

public void repeatIterationNoPlan()
                           throws Exception

Repeats the last iteration of the agent loop. This method will cause the agent to abort the current step of the agent loop and to go back to the first step of the loop. The current time step is not increased. A dialog will be displayed, saying that no plans were returned by the inference engine.

repeatIterationNoModel

public void repeatIterationNoModel()
                            throws Exception

Repeats the last iteration of the agent loop. This method will cause the agent to abort the current step of the agent loop and to go back to the first step of the loop. The current time step is not increased. A dialog will be displayed, saying that no models were returned by the inference engine.

repeatIteration

public void repeatIteration()
                     throws RepeatIterationException

Repeats the last iteration of the agent loop. This method will cause the agent to abort the current step of the agent loop and to go back to the first step of the loop. The current time step is not increased.

computeOneTrajectory

public Trajectory computeOneTrajectory(MathSet fileList,
                                       int currTime,
                                       int lastTime)
                                throws Exception

Computes one model of the given list of files, with the given values for A-Prolog parameters currtime and lasttime, and returns the Trajectory contained in the model.

If an error occurs during the execution of the inference engine, repeatIteration() is invoked.

Parameters:

fileList - the list of file names to be used by the inference engine

currTime - the value of A-Prolog parameter currtime

lastTime - the value of A-Prolog parameter lasttime

Returns:

the Trajectory contained in the model

computeAllTrajectories

public TrajectorySet computeAllTrajectories(MathSet fileList,
                                            int currTime,
                                            int lastTime)
                                     throws Exception

Computes all the models of the given list of files, with the given values for A-Prolog parameters currtime and lasttime, and returns the TrajectorySet represented by the models.

If an error occurs during the execution of the inference engine, repeatIteration() is invoked.

Parameters:

fileList - the list of file names to be used by the inference engine

currTime - the value of A-Prolog parameter currtime

lastTime - the value of A-Prolog parameter lasttime

Returns:

the TrajectorySet represented by the models

agentLoop

public abstract void agentLoop()
                        throws Exception

The agent loop. This is the method that implementations of the agent loop must define.

run

public void run()

Invoked only by the APLAgent Manager

Overrides:

run in class Thread