S. Au and N. Parameswaran, The University of New South Wales, Australia
In this paper, we propose that in multiagent dynamic worlds, agents which are autonomous, need to be guided by attitudes for effective problem solving. Often agents, existing in dynamic and hostile worlds, are required to exhibit the specified problem solving behaviors for prolonged periods of time either continually or intermittently. In order to be able to perform these types of behaviors, autonomous agents need meta-level controls known as attitudes to guide them towards selecting the proper goals and actions. In this paper, we investigate the role of attitudes in problem solving in dynamic worlds, and suggest several attitudes for the agents in a hostile dynamic world, the fire world. We then evaluate and compare the problem solving behaviors of the agents in a simulated fire world using different types of attitudes. In a dynamic environment, meta-level control is needed not only to improve the efficiency of reasoning, but also the accuracy and utility of the results of reasoning. - Martha Pollack in The uses of plans, AI 57(1992) 43-68.