Wiki code for Examples

Hide Line numbers
1: $xwiki.ssx.use("XWiki.Lightbox")
2: $xwiki.jsx.use("XWiki.Lightbox")
3: 
4: 1 Examples
5: 
6: The Jadex software includes several small to medium size examples
7: demonstating different aspects of BDI agent programming. In the
8: following the most interesting of them are described shorty.
9: If you have [Java Webstart>http://java.sun.com/products/javawebstart/]
10: installed, you even can launch the examples directly from your browser.
11: 
12: 1.1 Hunter Prey
13: <table border="0" cellspacing="0" cellpadding="0">
14: <tbody>
15: <tr>
16: <td valign="center">
17: <p>
18: The hunter prey scenario consists of two kinds of creatures living in a grid world.
19: The basic task of hunters is to chase, while preys move around looking for food.
20: Both kinds of creatures have to act autonomously in the environment on basis of
21: their current local view, experiences made in the past and communications with
22: others. Besides hunter and preys the environment accomodates other passive world objects.
23: On the one hand there are trees on many squares that prohibit creatures running on
24: such fields and on the other hand little plants grow at random squares at the map.
25: These plants can be eaten by the preys if they are on the same field. The scenario
26: is round-based with a fixed time slot for each round. This means that all creatures
27: in the world have to issue their next action (moving to some adjacent square or
28: eating something on the current square) with that round time. If no action is
29: announced no action will be executed. The environment will decide in each round
30: if an action succeeds or fails.
31: </p>
32: </td>
33: <td valign="center" align="center" style="padding-left=10pt">
34: <form method="GET" action="http://jadex.informatik.uni-hamburg.de/docs/webstart/hunterprey.jnlp">
35: <a href="$xwiki.getAttachmentURL($doc.fullName, "hunterprey.png")" rel="lightbox" title="Hunter Prey"><img src="$xwiki.getAttachmentURL($doc.fullName, "hunterpreysmall.png")"
36: alt="Hunter/Prey" border="0" width="161"/></a>
37: <input value="Launch (2.33 MB)" type="submit"/>
38: </form>
39: </td>
40: </tr>
41: </tbody>
42: </table>
43: 
44: 
45: 1.1 BlackJack
46: <table border="0" cellspacing="0" cellpadding="0">
47: <tbody>
48: <tr>
49: <td valign="center">
50: <p>
51: This example allows for playing BlackJack with one or more computer opponents.
52: Basically, it consists of three different agent types: the manager, the dealer
53: and the player.
54: The Blackjack-Manager can be used to start the game.
55: It shows up with a user interface that can be used to start and stop the
56: local dealer as well as local players. The user may also customize the
57: players, i.e. give them a name and an initial account, choose a strategy
58: and assign a different color. When a player agent is started it will search
59: a dealer to play a game. After at least one player has registered at the
60: Dealer the dealer starts playing games. The player may stay at the table as
61: long as it owns money for playing.
62: </p>
63: </td>
64: <td valign="center" align="center" style="padding-left=10pt">
65: <form method="GET" action="http://jadex.informatik.uni-hamburg.de/docs/webstart/blackjack.jnlp">
66: <a href="$xwiki.getAttachmentURL($doc.fullName, "blackjack.png")" rel="lightbox" title="BlackJack"><img src="$xwiki.getAttachmentURL($doc.fullName, "blackjacksmall.png")"
67: alt="BlackJack" border="0" width="161"/></a>
68: <input value="Launch (5.67 MB)" type="submit"/>
69: </form>
70: </td>
71: </tr>
72: </tbody>
73: </table>
74: 
75: 
76: 
77: 1.1 Blocksworld
78: <table border="0" cellspacing="0" cellpadding="0">
79: <tbody>
80: <tr>
81: <td valign="center">
82: <p>
83: In the blocksworld coloured blocks are placed on a table as
84: towers of single blocks.
85: This example only consists of a single agent, which has an
86: internal representation of the blocksworld. When the agent is started, a
87: window is opened that shows the current state of the blocksworld,
88: and another panel where you can construct a desired target configuration.
89: When the "create goal" button is clicked, the agent tries to achieve the
90: target configuration by restacking blocks on the table. For this purpose
91: a simple solution strategy is applied. Basically, all bad blocks are put
92: onto the table and are stacked on towers that fit the target configuration.
93: </p>
94: </td>
95: <td valign="center" align="center" style="padding-left=10pt">
96: <form method="GET" action="http://jadex.informatik.uni-hamburg.de/docs/webstart/blocksworld.jnlp">
97: <a href="$xwiki.getAttachmentURL($doc.fullName, "blocksworld.png")" rel="lightbox" title="Blocksworld"><img src="$xwiki.getAttachmentURL($doc.fullName, "blocksworldsmall.png")"
98: alt="Blocksworld" border="0" width="161"/></a>
99: <input value="Launch (2.33 MB)" type="submit"/>
100: </form>
101: </td>
102: </tr>
103: </tbody>
104: </table>
105: 
106: 
107: 
108: 
109: 1.1 Cleanerworld
110: <table border="0" cellspacing="0" cellpadding="0">
111: <tbody>
112: <tr>
113: <td valign="center">
114: <p>
115: The cleanerworld is based on the idea that an autonomous cleaning robot has
116: the task to clean up dirt in some environment. In our scenario of the cleaner
117: world the main system objectives are to keep clean a building at day, e.g. a museum,
118: and to guard the building at night. To be more concise we think of a group of
119: cleaning robots that are located in the building and try to accomplish the overall
120: system goals by pursuing their own goals in coordination with other individuals.
121: Therefore, three key goals for an individual cleaning robot were identified.
122: First, it should clean its environment at day by removing dirt whenever possible.
123: The cleaning robot therefore has to pick-up any garbage and carry it to a near waste bin.
124: Secondly, it has to guard the building at night by performing patrols that should be
125: based on varying routes. Any suspicious occurrences that it recognises during its
126: patrols should be reported to some superordinated authority. Thirdly, it should keep
127: operational by monitoring its internal states such as the charge state of its battery or
128: recognised malfunctions. Whenever its battery state is low it has to move to the charging
129: station.
130: </p>
131: </td>
132: <td valign="center" align="center" style="padding-left=10pt">
133: <form method="GET" action="http://jadex.informatik.uni-hamburg.de/docs/webstart/cleanerworld.jnlp">
134: <a href="$xwiki.getAttachmentURL($doc.fullName, "cleanerworld.png")" rel="lightbox" title="Cleanerworld"><img src="$xwiki.getAttachmentURL($doc.fullName, "cleanerworldsmall.png")"
135: alt="Cleanerworld" border="0" width="161"/></a>
136: <input value="Launch (2.33 MB)" type="submit"/>
137: </form>
138: </td>
139: </tr>
140: </tbody>
141: </table>
142: 
143: 
144: 
145: 1.1 Garbage Collector
146: <table border="0" cellspacing="0" cellpadding="0">
147: <tbody>
148: <tr>
149: <td valign="center">
150: <p>
151: The garbage collector example is a modified and simplified version of the
152: cleaner world scenario. It consists of two kinds of agents in an environment
153: covered with garbage. Collector agents search for dirt and if found bring it
154: to one of the burner agents. The burner agents do not move. They pickup garbage
155: at their position and burn it. The environment in this example is represented
156: as simple grid world. All agents can only see things that are situated in their
157: currently taken position. For this reason the collectors simply run in a
158: predefined pattern through the environment and try to pickup garbage as soon
159: as they reach a dirty square. If a collector managed to pickup a piece
160: of garbage it heads directly to one of the garbage burners and lays down the
161: garbage. Thereafter it returns to the point it has found the waste and continues
162: its search for garbage from there.
163: </p>
164: </td>
165: <td valign="center" align="center" style="padding-left=10pt">
166: <form method="GET" action="http://jadex.informatik.uni-hamburg.de/docs/webstart/garbagecollector.jnlp">
167: <a href="$xwiki.getAttachmentURL($doc.fullName, "garbagecollector.png")" rel="lightbox" title="Garbage Collector"><img src="$xwiki.getAttachmentURL($doc.fullName, "garbagecollectorsmall.png")"
168: alt="Garbage Collector" border="0" width="161"/></a>
169: <input value="Launch (2.33 MB)" type="submit"/>
170: </form>
171: </td>
172: </tr>
173: </tbody>
174: </table>
175: 
176: 
177: 
178: 1.1 Marsworld
179: <table border="0" cellspacing="0" cellpadding="0">
180: <tbody>
181: <tr>
182: <td valign="center">
183: <p>
184: Several interacting agents have the task to explore the environment for ore
185: resources and bring as much ore as possible to the agents homebase. When the
186: mission time has expired the agents have to abort their current actions
187: and return to the homebase. The different agent types include sentry agents,
188: production agents and carry agents.
189: The sentry agent has the task to find ore resources inspect them if they can
190: be exploited. Therefore the sentry agent has the greatest vision of all agent
191: types. To find the ore resources more quickly all other agents report to the
192: sentry about resources they explored.
193: The production agent is called to a target from a sentry to produce as much
194: ore as the capacity of the resource permits. When finished the agents calls
195: for carry agents to bring the ore to the homebase.
196: The carry agent has the task to bring ore from targets to the homebase.
197: It is called by the production agent.
198: </p>
199: </td>
200: <td valign="center" align="center" style="padding-left=10pt">
201: <form method="GET" action="http://jadex.informatik.uni-hamburg.de/docs/webstart/marsworld.jnlp">
202: <a href="$xwiki.getAttachmentURL($doc.fullName, "marsworld.png")" rel="lightbox" title="Marsworld"><img src="$xwiki.getAttachmentURL($doc.fullName, "marsworldsmall.png")"
203: alt="Marsworld" border="0" width="161"/></a>
204: <input value="Launch (2.33 MB)" type="submit"/>
205: </form>
206: </td>
207: </tr>
208: </tbody>
209: </table>
210: 
211: 
212: 
213: 1.1 Puzzle
214: <table border="0" cellspacing="0" cellpadding="0">
215: <tbody>
216: <tr>
217: <td valign="center">
218: <p>
219: This example is adapted from the commercial agent
220: platform JACK(TM) from Agent Oriented Software.
221: The Jadex implementation is very similar to
222: allow performance measurements between both
223: platforms.
224: The example shows a puzzle game played by one agent.
225: It consists of a board with white and red
226: pieces. The objective is to switch the positions
227: of both kinds of pieces whereby the following rules
228: for making a move exist:
229: 
230: * white pieces move right or down to an adjacent free field.
231: * white pieces jump right or down over a red piece to a free field.
232: * red pieces can only move up or left with the same restrictions as white pieces.
233: * the color of a piece to move is not specified.
234: 
235: The agent uses meta-level reasoning to solve the
236: puzzle. It creates a goal to make a move and find
237: the solution. For this goal a plan for each
238: possible move is created. To decide which move
239: plan to test first a meta-level goal is created.
240: The choose move plan handles the meta goal and
241: decides according to a specified strategy.
242: </p>
243: </td>
244: <td valign="center" align="center" style="padding-left=10pt">
245: <form method="GET" action="http://jadex.informatik.uni-hamburg.de/docs/webstart/puzzle.jnlp">
246: <a href="$xwiki.getAttachmentURL($doc.fullName, "puzzle.png")" rel="lightbox" title="Puzzle"><img src="$xwiki.getAttachmentURL($doc.fullName, "puzzlesmall.png")"
247: alt="Puzzle" border="0" width="161"/></a>
248: <input value="Launch (2.33 MB)" type="submit"/>
249: </form>
250: </td>
251: 
252: </tr>
253: </tbody>
254: </table>
255:
Examples

Top Menu

Wiki code for Examples

Search

Navigation

About

Download

Internal

Usages

Resources

Links

Quick Links

Users Currently Online
