The impact of domain dynamics on intelligent robot design
journal contributionposted on 01.02.2003, 00:00 by Douglas A. Reece, Steven A. Shafer
Abstract: "Autonomous mobile robots should be able to perform useful tasks in complex and dynamic environments. In an attempt to achieve this goal, artificial intelligence and robotics research has given rise to a wide variety of intelligent systems. Only recently have these systems begun to address the issues involved in solving problems and planning action with limited resources in complex and dynamic domains. To this point, the literature has not provided an analysis of how these factors affect system architecture. This paper discusses how resource limitations and domain characteristics affect the basic structure of system architecture, and how this structure impacts the system's ability to solve cognitive problems. We set forth four general design issues for systems: modularization, parallelism, execution control, and communication.In each case, we explain how the demands of problem solving ability and real time response conflict, thereby making it difficult to integrate planning and control in the same system. With this understanding of the implications of design choices, it is possible to evaluate the strengths and weaknesses of real and potential architectures. The range of design choices is illustrated with examples of real planners, controllers, and other intelligent systems."