Expert system control of an agent-based mobile robot

A great deal of interest exists in the fields of intelligent systems and robotics. Particularly, autonomous robots have proven to be very useful. Autonomous mobile robots have many potential uses in endeavors academic, industrial, and military beyond their current embryonic stages. Many architectures have been attempted for mobile robots, with varying degrees of success.; This thesis explores the use of a rule-based expert system to control an agent-based mobile robot. A proof of concept prototype is developed named ESAgent using free or open source software. ESAgent is portable, scalable, and extensible. ESAgent uses Player/Stage to simulate a mobile robot, interfaced with a JADE agent implemented in Java using the Javaclient library. Another agent utilizes an instance of the JESS expert system in conjunction with custom rule sets to communicate with other agents and control the simulated robot. Several rule sets are created to demonstrate different behaviors, and multiple instances can be started for a combined effect. ESAgent shows that this architecture is feasible, and demonstrates simple behaviors in navigating a simulated robot in a two-dimensional world. Further work on the rule sets may yield additional utility from the design.