Motivational Quantities : State-based Control for Organized Agents

T. Wagner
Abstract:Control for agents situated in multi-agent systems is a comp lex problem. This is particularly true in hard, open, dynamic en vironments where resource, privacy, bandwidth, and comput ational limitations impose restrictions on the type of infor mation that agents may share and the control problem solving option s available to agents. The MQ or motivational quantitiesframework addresses these issues by evaluating candidate tasks b ed on the agent’s organizational context and by framing contro l as a local agent optimization problem that approximates the gl obal problem through the use of state and preference.
What problem does this paper attempt to address?