Heuristic Area Cost Estimation for Observational Coverage Schedulers

This paper presents a comparison of heuris-
tics used to estimate the amount of time it
would take for a spacecraft to image an area
using Boustrophedon decomposition (Choset
and Pignon 1998). Machine learning tech-
niques are used to characterize algorithmic
performance of coverage algorithms. It is
shown that an ordinary least-squares linear
model is among the most accurate in a set of
constant and linear order regression models
both in terms of memory consumption and
schedule duration. These are demonstrated
using the ASPEN planning system (Fukunaga
et al. 1997) on the Eagle Eye domain.