Using Logic Programming to Enhance High Level Strategic Robotic Decision Making
Herbert Au · 2006
The Artificially Intelligent roBOt (AIBO) created by Sony is programmed using a reactive intelligence model. This means that the robot will receive certain inputs, and react to these inputs based on pre-programmed outputs. If we can move towards a proactive model, which is able to receive inputs and react in such a way that it automatically selects the best output which aids in achieving its final goal, we can create a more intelligent robot soccer team which will use and create its own strategies and tactics during game play. By creating a Tic-Tac-Toe behaviour using the Tekkotsu [9] framework on the AIBO, two versions of Tic-Tac-Toe were implemented; a reactive module and a proactive module. These were run separately on the AIBO, and it was shown that both were able to calculate the next move in a short period of time, and both used approximately the same amount of memory. Although it was shown that it is possible to run a longer, ‘logical’ version of Tic-Tac-Toe on the AIBO, the implementation of the logic program is a very complicated process. Because of the AIBO’s specialized operating system, it is unable to handle file pointers and file manipulations. Thus, existing Prolog (and other logic programming language) interpreters were unable to be used. Instead, a simple C++ program which attempts to emulate a logic program, specifically designed for tic-tac-toe was used. This means that to implement a soccer strategy, we would require a few more projects, including writing a compiler which converts prolog source into C++ source, and another which consists of implementing a strategy for the robots to follow. Many more lower level kinematics which will need to be called by the higher level decision making processes will also be required.