Simple Bargaining Agents for Decentralized Market-Based Control

market-based control, economic agents, bargaining, autonomous agent Market-Based Control (MBC) is a resource allocation and control technique where multi-agent systems are built to resemble free-market economies. The aim is that MBC systems exhibit the same decentralization, robustness, and capacity for self-organization as do real economies. MBC systems are relevant to Artificial Intelligence (AI) and robotics in at least two ways: first, the agents in a MBC system need to be robot-like in their ability to autonomously coordinate perception and action in dynamic and uncertain environments that include other agents; second, MBC systems could be used as the control technologies for robots and other “intelligent” autonomous agents. We critically review a selection of MBC systems. We argue that the MBC systems reviewed here are either implicitly reliant on centralized knowledge, or require human operators and hence are not truly automatic. We identify a major issue in creating truly decentralized and automatic MBC systems: the need for the system's agents to be capable of bargaining behaviors. Following this, we briefly summarize our current results and ongoing work in creating multi-agent systems where each autonomous agent has the ability to bargain with other agents. We demonstrate that markets composed of such agents exhibit desirable behaviors, and that such agents could form the basis of truly decentralized MBC systems.