Prediction, observation and estimation in planning and control

Abstract

The central role of reasoning about change in planning and control is studied. Some of the basic insights of control theory regarding the connection between observability and controllability are reviewed, and how these insights should inform research in planning is considered. A decision-theoretic approach to planning that subscribes to a basic cycle of prediction, observation, estimation and action selection is proposed. Some of the practical issues relating to this approach and alternatives in the form of offline compilation of decision models and learning systems to circumvent the need for accurate predictive models are examined. Problems in mobile robotics that require integrating obstacle avoidance, position estimation, and path planning are used to illustrate the approach.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>