Enhancing the degree of autonomy by creating automated components within a multi-agent system framework

Abstract

Humans have used machines to do more with less physical control and/or interaction (by increasing the Level of Automation (LoA)). Engineers have been tasked with progressively automating functions previously completed manually. Industry continues promoting a mechanized workforce in order to minimise business variations in human performance and maximise their productivity. Researchers also continue to pursue this goal using Artificial Intelligence (AI), Computational Intelligence (CI) and Machine Intelligence (MI) techniques. This process of transforming cognitive functionality into machine actionable form has encountered bottlenecks, such as scalability and inconsistent performance. More recently, Multi-Agent Systems (MASs) have been employed to provide a flexible framework for research and development, because of their ability to easily embody system complexity. The proposed framework should facilitate the development of components, their interoperability, coordination and cooperation techniques needed to solve complex, realworld problems within a dangerous and/or dynamic environment. Based on the effort to seek or facilitate human-like decision making within machines, the lack of success indicates that further research is required. This chapter discusses one possible avenue. It involves future research, aimed at achieving a cognitive sub-system for use on-board platforms. The framework introduced in this chapter is aimed at embedding traditional heuristic knowledge and higher order cognitive computation techniques. This conceptual mechanism could ultimately be used to implement a virtual mind when technology facilitates a real-time implementation and could be used to progressively enhance automation, achieve greater independence and enable more autonomous behaviour within control systems.