An application of genetic algorithms in smoothing automotive body surface

In car body design, reflection lines of automotive body surface have been used to identify and remove the bump or irregular curves on the outer skin surfaces of car body. Traditional approaches to correcting such reflection lines involve tedious manual work and expensive computation. They also rely heavily on designers' experiences. This paper explores how to use genetic algorithms to correct the reflection lines under user-specified constraints. This approach has the potential to alleviate these drawbacks. A method of measuring fitness is also devised which indicates how well the reflection lines comply with physical and aesthetic constraints. Two encoding schemes are proposed and experiments based on these schemes are analysed.