Development of Methodology for Programming Autonomous Agents

Abstract

A brief report discusses the rationale for, and the development of, a methodology for generating computer code for autonomous-agent-based systems. The methodology is characterized as enabling an increase in the reusability of the generated code among and within such systems, thereby making it possible to reduce the time and cost of development of the systems. The methodology is also characterized as enabling reduction of the incidence of those software errors that are attributable to the human failure to anticipate distributed behaviors caused by the software. A major conceptual problem said to be addressed in the development of the methodology was that of how to efficiently describe the interfaces between several layers of agent composition by use of a language that is both familiar to engineers and descriptive enough to describe such interfaces unambivalently