Methodological Assumptions Behind the Basic Agent Architecture

• Modularity

  The basic agent architecture aims to integrate the knowledge of separate AI disciplines into a single system. A planner module is included to provide the planning capability. To provide natural language capabilities, modules for text interpretation and generator are included. The two fields developed independently for the most part, but the basic agent incorporates the work of both in its architecture. It remains to be seen whether this modular technique can be extended to other capabilities, and even for these two capabilities, efficiency can be an issue.

• Simulation

  Homer does not actually perform sensing or manipulation in the real world. However, it does use a simulator, in which it can perform symbolic sensing and manipulation, assumed to be perfect. In a sense, this is part of the modularity assumption, in that sensing and manipulation modules could be added to the architecture to permit interaction with a real environment.