PROGRESSIVE GOAL-DIRECTED REASONING FOR REAL-TIME AI SYSTEMS

Real-time performance is increasingly attracting the attention of the artificial intellience (AI) community. In spite of their success in va rious applications classical AI systems are unsatisfactory when applie d to real-time applications. In this paper, we point out the problems raised by integration of real-time capabilities into multi-agent archi tectures. We present first the main requirements for real-time applica tions. Then we describe rite ability of goal-directed reasoning to add ress the reactivity and guarantee response rime in RT-SOS (Real-Time S ociety Of Specialists) in which asynchronous acquisition, interruptibl e reasoning, progressive reasoning and adaptive reasoning contribute t o meet the real-time requirements.