Introduction to Iris.3

The mobile robot group is dedicated to producing a autonomous robot capable of playing tic-tac-toe against a human opponent. The mobile robot (mobot) will include four major systems in order to achieve this goal. These include the communications system, the vision system, the AI system, and the hardware system. The primary disciplines involved in the Iris3 Mobot group are Computer Science and Industrial Technology.

The communication system controls the allocation of all hardware (arms, camera) and software (tic-tac-toe engine, vision system) devices, and serves as an interface for programs to use these devices. The mobot group has adapted SIE (Shelley Integrated Environment) from Illinois Wesleyan's Shelley Research Group to this purpose. For complete information on SIE, please see the Mind Project's version of the SIE Documentation. While SIE suits our purposes right now, in the future we may need to develop a custom system to coordinate communication. In order to use SIE, existing software had to be adapted to work with this interface. This consisted primarily of a program that played tic-tac-toe. Integration of this program is largely done, with some minor bugs that need to be resolved.

The primary sensor for the Tic-Tac-Toe project is a Connectix camera. We have been working on developing an interface to this camera that will suit our need, and nearly have it up and running. The camera will send images to the vision system, which is incomplete. Some of the tasks that remain on the vision system include the following: An artificial neural network (ANN) package must be chosen. Training examples must be developed (we have some examples, which may or may not be suitable). And the network must be trained to recognize the X's and O's of the playing field. Training is dependent on the camera system, which is still under development but nearing completion.

The AI system for the TTT robot is a TTT engine, which has been previously developed. Also needed is a controlling agent program that will control the activities of all other programs and resources (vision, the TTT engine are all considered "devices" in our communication model. An agent is needed to tell the devices what to do).

The effector is a single arm, which is controlled by a series of servo motors. These servo motors are, in turn, controlled by a Jameco controller card. The Jameco controller interfaces to a computer via a 9-pin serial connection. Scripts to control the movement of the arm around the Tic-Tac-Toe board are also needed.

Undergraduate researchers

  • Matt Eisenbraun (Applied Computer Science)

  • Andy Schmidgall (Applied Computer Science)

  • Wenyan Dong (Applied Computer Science)

  • Geoff Cluts (Applied Computer Science)

  • Jeremy Tessendorf (Industrial Technology)

  • Devin Izdepski (Applied Computer Science)

  • Mike Wiegand (Applied Computer Science), Wiegand@GTE.net

Faculty advisors