Anticipating computational demands when solving time-critical decision-making problems

摘要

This paper is concerned with anticipating computational demands to assist in allocating processing time to decision-making procedures in time-critical applications. The mechanisms for anticipating computational demands are discussed in Section 2 and a particular approach to time-critical decision making called response planning is presented in Section 3. This approach focuses on solving time-critical decision-making problems in dynamic environments with large state spaces and resource limitations. Section 4 relates our response planning approach to the general idea of anticipating computational demands. In Section 5 we describe an example traffic-control problem for multiple robot vehicles that demonstrates the applicability of the response planning approach to solving time-critical problems. A general overview of this work motivated by an air traffic control problem is found in [10].