Dynamic Evaluation of Coordination Mechanisms for Autonomous Agents

Abstract. This paper presents a formal framework within which autonomous agents can dynamically select and apply different mechanisms to coordinate their interactions with one another. Agents use the task attributes and environmental conditions to evaluate which mechanism maximises their expected utility. Different agent types can be characterised by their willingness to cooperate and the relative value they place on short- vs long-term rewards. Our results demonstrate the viability of empowering agents in this way and show the quantitative benefits that agents accrue from being given the flexibility to control how they coordinate.

• Type: Conference
• Year: 2001
• Where: EPIA
• Authors: Rachel A. Bourne, Karen Shoop, Nicholas R. Jennings
• Keywords: Artificial Intelligence, Autonomous Agents, Different Agent Types, EPIA 2001, Vs Long-term Rewards