User Manual 1.0 - MSE - Kansas State University
Contents
1. Page 7 of 11 GMoDS based Runtime Agent Role Interpreter User Manual 1 0 4 2 4 Advanced Movement 4 2 5 4 2 6 The AdvancedMovement capability wraps the more primitive RobotMovement and GPSImpl and extends them with advanced mapping and route planning capabilities AdvancedMovement is used to determine the agent s current location set goal movement locations plan routes between locations and then actually execute those routes Available Methods getCurrentLocation LocationData getDestinationLcation LocationData getOldLocation LocationData isAtDestinationLocation boolean move boolean setDestinationLocation loc LocationData Figure 8 Advanced Movement Methods Advanced Sensors The AdvancedSensors capability internally wraps the Breeze Smell and Sparkle sensors It also encapsulates the sonar capability It is responsible for aggregating sensor input from a variety of sources interpreting it and then storing it in the agent s map database It is also responsible for sharing this map information with other agents and merging their map data with the agent s map data Available Methods isGoldAtLocation loc LocationData boolean isLocationSearchable loc LocationData boolean isWumpiAtLocation loc LocationData boolean isAtDestinationLocation boolean readOthersMaps readSensors bo2. Page 5 of 11 GMoDS based Runtime Agent Role Interpreter User Manual 1 0 4 Designing Models The GMoDS based Runtime Agent Role Interpreter is designed to modular and extensible To achieve that aim new Role Level Goal Models Capabilities and Goal Capability Maps can be supplied by users to enable them to use the Role Interpreter in any multiagent system that they choose 4 1 Creating New Role Level Goal Models New Role Level Goal Models can be created to specify new behavior for agents to perform while playing the role the model defines Role Level Goal Models are simply regular GMoDS Goal Models whose root goal is a leaf level goal in the top level system Goal Model The leaf level goals of a Role Level Goal Model map directly to capability methods that the agent possesses Role Level Goal Models can be constructed using the agentTool3 Diagram Editor just like any other GMoDS goal model Role Level Goal Models are no different from standard GMoDS Goal Models except for the following restrictions 1 The number and names of leaf goal specification parameters must match with number and names of formal parameters of the methods to which they are to be mapped If they do not match runtime exceptions will be thrown when the Goal Capability Map is parsed at startup 2 Special true and false prefixed event names must be used if conditional execution and by extension iteration is desired These events only possess speci
3. Existing Capabilities The following existing capabilities are present in the system and may be used within Role Level Goal Models and mapped to by Goal Capability Maps 4 2 1 Advanced Bazooka The AdvancedBazooka capability extends the existing Bazooka capability with targeting capabilities and ammo counting Available Methods fire loc LocationData boolean getNextFiringLocation LocationData hasAmmo boolean hasFiringLocation boolean setTarget loc LocationData Figure 5 Advanced Bazooka Methods 4 2 2 Advanced Communication The AdvancedCommunication capability extends the existing CommunicationsImpl capability This capability can send or receive messages on specific channels or send and receive broadcast messages Available Methods broadcast channelID String content Object boolean broadcastIncludeSelf channelID String content Object boolean receive Boolean removeChannel channelID String boolean send agentID Uniqueldentifier channelID String content Object Boolean Figure 6 Advanced Communication Methods 4 2 3 Advanced Gold Grabber The AdvancedGoldGrabber capability extends the existing GoldGrabber capability This capability can grab gold report if it has gold and release gold Available Methods grab boolean putGoldIntoBin boolean hasGold boolean Figure 7 Advanced Gold Grabber Methods
4. Maps Creating new Goal Capability Maps is simple and straightforward When you have finished creating a new Role Level Goal Model and any new capability classes check all leaf level goals and see if they already contain a mapping to capability method If they do not then you will need to provide such a mapping If any mappings are missing a runtime exception will be thrown indicating which mappings need to be established The sample XML file GoalCapabilityMap xml provides mappings for all existing capabilities to goal names used by the Role Level Goals Models in within the project models subdirectory If you wish to use different leaf level goal names to map to existing capability methods either rename the existing mapping s goal_name attribute or provide a new mapping with a different goal_name attribute Similarly if you create a new capability class you will need to construct mappings for each public method that you with to use from your RoleLevelGoalModels lt AdvancedGoldGrabber gt lt Capability class AdvancedGoldGrabber package edu ksu cis macr simulator capabilities gt lt Method goal_name grab declared_name grab gt lt Method goal_name putGoldIntoBin declared_name putGoldIntoBin gt lt Method goal_name hasGold declared_name hasGold gt lt Capability gt 4 4 1 Figure 12 Example Goal Capability Mapping Goal Capability Map Schema For easy reference the Goal Capability Map Schema
5. contains information about how to obtain the interpreter s source code and executable binary It also contains information about how to setup and run the Agent Architecture Demo that shows the Role Interpreter in action Finally it contains information about how to design your own Role Level Goal Models Capabilities and Goal Capability Maps Installation Prerequisites The GMoDS based Runtime Agent Role Interpreter requires the following prerequisites in order to run e Java Runtime Environment 6 0 or higher Required Files To obtain the GMoDS based Runtime Agent Role Interpreter download the final GAA source archive from http people cis ksu edu kylhill phase_3 gaa zip The only required files from within the archive are gaa jar and the contents of the configs models and scenarios directories The other files within the archive are the project source code and documentation Alternatively the GAA project and all of its dependencies can be checked out from the K State CIS projects CVS server at projects cis ksu edu cvsroot gaa The setup and use of this CVS repository are outside the scope of this document Setup Extract the final GAA source archive gaa zip to the desired location on your computer No other installation steps are required Running the Agent Architecture Demo The GMoDS based Runtime Agent Role Interpreter contains an agent architecture demonstration in addition to its basic Role Interpreter functionality T
6. is provided below to aid in the construction of valid mappings To check your Goal Capability Map simply validate your XML file with the schema provided below This will guarantee that the file will at least be parsed correctly when given the Role Interpreter Page 9 of 11 GMoDS based Runtime Agent Role Interpreter User Manual 1 0 e GoalCapabilityMap GoalCapabilityMapType ceo f Capability 1 CapabilityType Figure 13 GoalCapabilityMap e Capability Ea CapabilityType package string class string s de Method 1 MethodType Figure 14 Capability Method MethodType goal_name string declared_name string ves df Parameter 0 ParameterType Figure 15 Method ParameterType package string name string class string Figure 16 Parameter lt xml version 1 0 encoding UTF 8 gt lt xs schema xmlns xs http www w3 org 2001 XMLSchema gt lt xs element name Parameter gt lt xs complexType gt lt xs attribute name package type xs string use required gt lt xs attribute name name type xs string use required gt lt xs attribute name class type xs string use required gt lt xs complexType gt lt xs element gt lt xs element name Method gt lt xs complexType gt lt xs sequence gt lt xs element ref Parameter minOccurs maxOccurs unbounded gt lt xs sequence gt lt xs attribute name goal_name type xs
7. world and kill all of the Wumpi monsters that guard it While carrying out these tasks agents represented by the robot icons in the lower left must avoid falling into pit traps or being attacked by the Wumpi monsters Agents are scored based on how efficiently they search the map how many Wumpi are killed and how much gold is retrieved Page 4 of 11 GMoDS based Runtime Agent Role Interpreter Figure 3 Agent Map Display User Manual 1 0 Several other windows similar to the one shown in Figure 3 should show up as well These windows show what each agent thinks the world around them looks like The key for this map is given in Figure 4 Tile Color Meaning White Searched Tile Nothing Special Light Gray Unsearched Tile Dark Gray Unknown Obstacle Black Wall Red Wumpi Light Orange Possible Pit Dark Orange Possible Wumpi Danger Yellow Possible Gold Green Claimed Goal Tile Blue Agent Active Magenta Agent All Goals Complete Figure 4 Agent Map Display Key All agent activity within this scenario is controlled by the Role Level Goal Models supplied within the models subdirectory of the archive These goal models provide the instructions that agents are to carry out as they pursue their assigned goals The mappings of these goals to agent capability methods are defined by the Goal Capability Map The design and construction of these models is discussed in section 4
8. GMoDS based Runtime Agent Role Interpreter User Manual 1 0 User Manual For a GMoDS based Runtime Agent Role Interpreter Version 1 0 Submitted in partial fulfillment of the requirements of the degree of MSE Kyle Hill CIS 895 MSE Project Kansas State University Page 1 of 11 GMoDS based Runtime Agent Role Interpreter User Manual 1 0 Table of Contents t troduction ostia ti 3 2 Installatio Nenana e a re 3 Del a Prerequisite Sian eeni E eaaa 3 2 27 Reg ired A it chats nod 3 LS CSEE AN EE A RE ake ea hs 3 3 Running the Agent Architecture Demo s ccs seclesssaveyecsassgicsesssrconscaqnes yes cgasdaseahedacasacedevagnennenes 3 4 Designing Models ra 6 4 1 Creating New Role Level Goal Models ooonocccnnociconoccnoncccnonccconnncconncccnnncncnnncncnnncnnnnnnnos 6 4 27 Exisunge Capabilities den a a E EE E a N TONS 7 421 Advanced Bazooka sio en e a LE aden ees enc E E Sea 7 4 2 2 Advanced A A RR anes es 7 4 2 3 Advanced Gold Grabber diiniita a 7 ADA Advanced MONEDERO dias 8 ADS Advanced SONSOTS usina it iaa 8 ADO gt OIEA Sainte o RT SO NAS 8 A e E 9 4 4 Creating New Goal Capability Maps ii 9 4 4 1 Goal Capability Map Schema oooococnnncccnoncccnononononanonnnanononanononononnnnnnnnnnnnonnn cnn ncnnonnnno 9 Page 2 of 11 GMoDS based Runtime Agent Role Interpreter User Manual 1 0 2 2 1 2 2 2 3 3 Introduction This document serves a user manual for the GMoDS based Runtime Agent Role Interpreter It
9. al meaning if they originate from a leaf level goal whose mapped method returns a boolean value If they are used on a non boolean method then they have no special meaning If the method returns true at runtime only the true prefixed events are executed Likewise if the method returns false at runtime only the false prefixed events are executed 3 Only one instance goal in a triggers cycle may be active at a time This is due to the way GMoDS determines how specification goals are completed A custom patch was applied to GMoDS for this project to allow execute of triggers cycles under most cases This only means that Role Level Goal Models cannot support parallelism within a Role if conditional execution is also required Additional work on GMoDS could remove this restriction 4 At most only one new specification parameter may appear in an event leaving a given goal It is assumed that this specification parameter represents the return value from the invoked method Any and all previous instance and inherited parameters can be passed along as part of the event without any restrictions There is no real way for the system to deterministically provide values for more than one new specification parameter If this behavior is required then leaf level goals should be split up into smaller ones that only generate a single new event parameter when executed Page 6 of 11 GMoDS based Runtime Agent Role Interpreter User Manual 1 0 4 2
10. o run this demonstration execute the following command on a command line from the directory where you extracted the project archive java jar gaa jar scenarios WumpiWorldFull xml configs GoalCapabilityMap xml Figure 1 Sample Demonstration Invocation The first parameter to gaa jar provides the path to the CROS environment description XML file This file provides information describing the environment the agents will exist within Page 3 of 11 GMoDS based Runtime Agent Role Interpreter User Manual 1 0 as well as their starting locations and other parameters Several sample scenarios have been provided within the scenarios subdirectory The second parameter to gaa jar provides the path the Goal Capability Map description XML file This file provides information to map specification goal and parameter names from Role Level Goal Models to actual capability method calls at runtime Instructions for producing your own Role Level Goal Models and Goal Capability Maps are given in section 4 GiWmpiWoid O O O File Control View Did Hak ja Figure 2 Agent Architecture Demonstration A window similar to the one shown in Figure 2 should appear after invoking the given command This window is the standard a CROS environment display for the Wumpi World scenario This scenario was given as a final project to students in CIS 844 during the fall of 2010 The goal of this scenario is to gather all of the gold in the
11. olean sendOthersMaps setLocationUnsearchable loc LocationData Figure 9 Advanced Sensors Methods OMACS Interface The OmacsInterface is a special capability that allows Role Level Goal Models to communicate with the rest of the OMACS system It currently only allows for the construction and assignment of a single ReturnGoldGoal However in the future it could be extended to provide generic OMACS interface capabilities such as role assignment failure reporting and reorganization Available Methods createReturnGoldGoal Figure 10 OMACS Interface Methods Page 8 of 11 GMoDS based Runtime Agent Role Interpreter User Manual 1 0 4 3 Creating New Capabilities New capabilities can be added to the system very easily To make a capability that can be used within a Role Level Goal Model and Goal Capability Map simply state that the capability implements the empty GaaCapability interface and then annotate all parameters of all methods that you wish to export to the Goal Capability Map with the custom Name annotation The name you give the parameter is the name that must be called out in the Goal Capability Map As sample annotated method signature is provided below for reference public final boolean addChannel Name channelID final String channelID Name channel final CommunicationChannel channel 4 4 Figure 11 Example Method Annotation Creating New Goal Capability
12. string use required gt lt xs attribute name decLared_name type xs string use required gt lt xs complexType gt lt xs element gt lt xs element name GoalCapabilityMap gt lt xs complexType gt lt xs sequence gt Page 10 of 11 GMoDS based Runtime Agent Role Interpreter User Manual 1 0 lt xs element ref Capability maxOccurs unbounded gt lt xs sequence gt lt xs complexType gt lt xs element gt lt xs element name Capability gt lt xs complexType gt lt xs sequence gt lt xs element ref Method maxOccurs unbounded gt lt xs sequence gt lt xs attribute name package type xs string use required gt lt xs attribute name class type xs string use required gt lt xs complexType gt lt xs element gt lt xs schema gt Figure 17 The Goal Capability Map Schema Page 11 of 11
Download Pdf Manuals
Related Search