Toward Real-World Science Mission Planning

In this paper we describe the use of our planning and scheduling framework, HSTS, to support science mission planning. This work is part of an overall project to enable a small team of scientists to control the operations of a spacecraft. The present process is essentially carried out by human operators. Users (scientists and operators) rely on a non-codified understanding of the different spacecraft subsystems and of their operating constraints. They use a variety of software tools to support their decision making process. A successful use of AI planning and scheduling technology must consider the types of decision making that need to be supported/automated, the nature of the domain constraints and the capabilities needed to address them successfully, and the nature of external software systems with which the core planning/scheduling engine needs to interact.