Managing Variable and Cooperative Time Behavior

Abstract

This paper is about the requirements and architectural considerations that provide a SORT system with processes for observing, modeling, simulating, predicting, deciding, and acting in an external environment. For our purposes, ``real time'' (RT) means coordinated with an external source of time or with sequences of events over which the system has no direct control. It is this unpredictability in the timing of responses that is the hardest constraint on a real-time system design, especially when it is known a priori that the system cannot keep up with all important events, and that ``as fast as possible'' is not appropriate for some external interactions. We will describe a testbed that we are developing as a student team project at California State Polytechnic University, Pomona (Cal Poly Pomona) to experiment with SORT strategies, and a set of games that we will use to benchmark performance. Then we will describe some useful technical background from several areas: reasoning and representation processes, situation theory, levels of meaningfulness in knowledge, and activity loops. Finally, we show how these concepts apply to SORT agent knowledge and coordination. Our contribution here is to outline a set of problems (in the form of cooperative games) that we hope others in the community will adopt as one method for benchmarking models, methods, strategies, and other processes used in SORT systems.