Proficiency-Driven Decision-Making for Networks of Autonomous Agents

Abstract

Autonomous agents play a crucial role in various modern fields, including emergency response and urban security. Their ability to operate effectively without direct human supervision is essential, especially in high-stakes situations. A key challenge is enabling these agents to evaluate their proficiency in completing tasks and use this evaluation for informed decisionmaking. This paper explores the use of metric based on the assessment of autonomous agents' proficiency and applies it to improve their decision-making at run time. In this context, proficiency self-assessment will improve agent navigation, enabling agents to more effectively complete their mission tasks, such as reaching a destination area and enhancing estimation accuracy.