A Layered Adjustable Autonomy Approach for Dynamic Autonomy Distribution

摘要

Autonomy adjustment to a system requires a mechanism to implement the roles of the autonomous entities in the system. The required degree of autonomy to which the autonomous entities adhere is a highly debated topic. On one hand, people argue that strict minimal autonomy to the autonomous entities is sufficient in producing reliable systems. On the other hand, others deliberate that the entities with full autonomic capabilities are essential aspects of advanced intelligent and flexible systems. The adjustable autonomic agent approach can be a solution for both cases. In this paper, we extend the idea of modeling a spectrum of autonomy in a layered structure, where the agents can act at different layers of autonomy in order to fulfill the system's autonomic conditions. Consequently, a logical representation of the conceptual model of Layered Adjustable Autonomy (LAA) is proposed. The LAA model aims to give the system implicit control over the agents' decisions whenever necessary by managing the agents' autonomy, ensuring quality and robust decision-making. An Autonomy Analysis (AAM) and Situation Awareness (SAM) Modules are proposed to attest the dynamic distribution of agent autonomic performance to a degree of autonomy level.