CSE 6362 Intelligent Environments

Final Project

Project Proposal due by midnight on Friday, April 5, 2002.
Project Demonstration in class on Monday, April 29, 2002.
Final Project and team evaluations due by midnight Monday, May 6, 2002.

Project Overview

• Agent Behavior. You should first define a set of behaviors that your agent will exhibit in the environmenent. The intelligent environment will monitor these bahaviors through sensors, predict the agent's actions, and make decisions to modify the environment in order to facilitate the agent's behaviors. These behaviors should be expressed as a Markov model, where each node in the model is a description of the state of the environment. Transitions between nodes, and their associated probabilities, define the possible behaviors of the agent.

• Sensors. You should use at least three different types of sensors to monitor your agent's behavior as they act in the environment. You should describe what type of sensors you are using, where they will be placed, and how their data will be transmitted and stored. Each sensor will have a CORBA server and will simulate the generation of values appropriate to the agent's behavior. Note that some of the "sensors" may actually represent aspects of the environment to control (e.g., lights). These too will have a CORBA server that accepts control commands from the decision maker.

• Network. You should describe the networking infrastructure of the environment to support your scenario. This part of the project will not be implemented, just described in the documentation. However, the network infrastructure should be appropriate for the demands of the environment. All communication in the intelligent environment will be made between CORBA objects.

• Database. You should describe the database infrastructure of the environment to support your scenario. You do not have to use a complete database system to store information (although some data storage mechanism will be necessary), but such use will increase the difficulty score of your project. In either case, you should describe the database infrastructure and how it is appropriate for the demands of the environment. The database component should be a CORBA object that collects sensor information and provides data as requested from the prediction and decision-making objects.

• Prediction. You should implement one or more prediction algorithms capable of predicting your agent's actions from the collected sensory information. The prediction component will be implemented as one or more CORBA objects that request data from the database object and either send the predictions back to the database or provide them on request from the decision maker.

• Decision-Making. You should implement a decision-making object that decides what actions the intelligent environment should take at each moment in time. The actions will generate CORBA requests to the appropriate "sensor" or effector to perform the action.

• Integration. The above components will be integrated via CORBA-based communication. A GUI should also be provided that allows start/stop control over the environment and displays the status of the environment.

Project Proposal

On or before the above Project Proposal due date, you should submit to me (holder@cse.uta.edu) a proposal for your project outlining the behaviors, sensors, predictions, and decisions to be implemented and used and how the entire system will be demonstrated. Project proposals must be approved by the instructor.

Project Demonstration

You will demonstrate your project in class on the date indicated on the schedule by first briefly describing your implementation of the above components and then visually demonstrating the adaptability of the environment using your GUI. The demonstration must show the environment adapting to the agent in some way by learning to perform some appropriate action.

Final Project

Your final project will consist of a project report, presentation, and all source code. The report should describe in detail all the above components: agent behaviors, sensors, network, database, prediction, decision-making, and integration. The report should also describe the demonstration and give precise instructions for running the demonstration. The presentation should be a set of PowerPoint slides giving an overview of your project, as presented at the demonstration. The source code should include everything needed to run your scenario. Your source code should follow the Coding Standard and use a Makefile to support easy compilation. Package up everything according to Assignment Turn-In Information as appropriate for the project specification and email to holder@cse.uta.edu by the above final project deadline.

Grading

• Agent behavior
• Sensor component
• Network component
• Database component
• Prediction component
• Decision-making component
• Integration
• Demonstration
• Documentation
• Source code
• Difficulty