|
This paper describes the implementation ofa mission management system for a mobile robot. It uses the concepts of perception, temporal, and spatial planning as key components in a system which must reason about route planning, task planning, and self-localisation through the use of on-board sensors. Other planning considerations include the rate at which resources, such as fuel, are consumed and how they might be replenished. Together, these constitute a multiple constraint planning problem for which an optimal solution may be very hard to find. However, sub-optimal "feasible" solutions may be acceptable, especially in a situation where a human operator is involved in setting the robot mission, and the planning system is acting as a decision aid. Our research into this problem has resulted in the development of an Object Oriented "Intelligent System Controller" (ISYC). This is implemented in C++ and supports the preparation of a robot mission, its execution and monitoring, and any re-planning which might be required as a result of a planning failure being revealed during a mission (for example, finding a route segment to be completely blocked, arriving late at a given location, failing to maintain sufficient selflocalisation accuracy to complete a mission, or running low on fuel). A planning algorithm based on a guided depth-first search through a weighted graph is used to implement much of the ISYC. Bayesian decision analysis is used as the basis of the guiding. We begin by presenting some background to this work, including a brief description of some architectures for the control of intelligent mobile robots. The ISYC architecture used in our work is then described in some detail. Next, the multiple constraint planning problem is introduced, and our approach to solving it is discussed. Finally, we describe the application of the resulting intelligent control system to the preparation and monitoring of missions for a material transport robot, and present some conclusions and some suggestions for further research.
|
