AAAI Conference 1994 Short Paper
Local Search in the Coordination of Intelligent Agents
- Daniel E. Damouth
In a world inhabited by numerous agents pursuing distinct goals, conflicts are inevitable. To succeed in the environment, an agent must explicitly reason about the behaviors of other agents as well as itself, and be prepared to find new behaviors that are more coordinated. Because traditional AI has had great success viewing problem solving as a search in a problem space, we have chosen to represent the process of coordination as a distributed search (Durfee et al. 1994). In searching through a joint behavior space for coherent coordination patterns, an agent must observe three kinds of constraints: its abilities, its goals, and the activities of other agents in the environment.