Multiagent Systems for Exploring an Environment

In this homework, you will learn how to create agents with two different kinds of internal reasoning architectures. You will use JESS to create rule-based agents and JADEX to create BDI-based agents. Each multiagent system will have the same task: to explore an environment. Assume that the agents live in a two-dimensional grid world that contains a treasure, such as shown below. Also assume that each agent can move one step in any of the eight directions possible from a given square. As the agents explore their environment, they will report to each other about what they find in each square they visit. Create at least three agents and try a grid of size 10x10.

Part 1: Rule-Based (JESS) Agents

Create at least three JESS-based JADE agents. When each agent is created, it will be located randomly within the environment. Each agent will know the size of the environment and where it is located within it. You can implement the environment as either a simple agent or as a global data structure accessible to all of the agents.

When an exploring agent is created, it will store its location in the environment structure (or inform the agent representing the environment about its location) and find out the names of all other exploring agents. It needs the names in order to communicate with them.

Each agent will then behave as follows:

It will wake up.
It will move randomly to one of its eight adjacent squares, except that it will not overstep a boundary and it will not visit a square that it or any other agent has already visited.
It will update the environment with its new location and query the environment to find out if the square contains the treasure.
It will report to the other agents what it has found in that square, using FIPA-standard messages.
It will destroy itself when the treasure is found.
It will go to sleep for 5000 milliseconds.
It will repeat the above actions.

I suggest that you write the rules for an agent as a .clp file and test them using CLIPS or JESS, before you try to make them part of a JADE agent. I also suggest that you develop this incrementally, i.e., develop the environment, develop an agent that can interact with the environment, develop an agent that can communicate with another agent, etc.

Note that each agent can at any time receive and process messages from other agents about what they have found. It should have a separate behavior for this, so that it can handle this concurrently with its exploration activities. Also, when it receives a message from another agent, it will add that agent's name to its list of agents with which it communicates, if the name is not already there (this will enable new agents to be added to the exploration team at any time).

Either print in a console window the location of the treasure and the locations of each agent and what they find in that location OR draw a graphical window showing the agents and the treasure.

Part 2: BDI-Based (JADEX) Agents

Create at least three JADEX agents. The agents will behave exactly like the JESS-based agents, except that they will maintain their knowledge in the form of beliefs, desires, and intentions. The agents will have beliefs about the contents of the squares they visit or hear about. The agents will have a desire to find the treasure. The agents will intend to explore their environment until the treasure is found. The agents will truthfully inform the other agents what they have discovered so far.

Deliverables

The code for the two multiagent systems.
A text file containing the JESS rules that control each agent's behavior.
A text file containing the beliefs, desires, and intentions of the JADEX agents.
A README.txt file containing your name and the instructions for executing your systems.
Upload your deliverables to the dropbox by Midnight on Sunday May 4.