INTELLIGENT AGENT ARCHITECTURES : Reactive Planning Testbed Final Technical Report

Much recent work has addressed the problems of uncertain, real-time domains. This paperargues that these negative characterizations of domains are inadequate: they make an alreadydifficult problem harder. Focussing on these characteristics leads one into intractable searchproblems and unrealistic noise models. We need rather to focus on the positive attributesof domains that make action possible. I analyze the success of Sonja and Pengi in these-'-terms. I propose that research should seek new facilitating features and architectures thatcan exploit them. This research was supported by the Air Force Office of Scientific Research under contract F49620-89-C-0055. Copyright @ Da,vid Chapman 1990. All rights reserved.