Robust local search for
spacecraft operations using adaptive noise
|
|||||||
Abstract Randomization is a standard
technique for improving the performance
of local search algorithms for constraint satisfaction.
However, it is well-known that local search algorithms
are sensitive to the noise values selected. We investigate the
use of an adaptive noise mechanism in an iterative repairbased
planner/scheduler for spacecraft operations. Preliminary
results indicate that adaptive noise makes the use of randomized
repair moves safe and robust; that is, using adaptive
noise makes it possible to consistently achieve performance
comparable with the best tuned noise setting without the need
for manually tuning the noise parameter.
|
|||||||
pdf file ![]() |