作者: Auke Ijspeert , David Willshaw , John Hallam , Jimmy Or
DOI:
关键词:
摘要: This paper investigates the evolutionary design of efficient connectionist swimming controllers for a simulated lamprey. Efficiency is defined as ratio forward speed to backward mechanical wave speed.Using lamprey model proposed by Ekeberg (1993) and extending work Ijspeert et al. (1999) on evolving central pattern generators (CPGs) through genetic algorithms (GAs), we investigate space possible neural configurations which satisfies property high efficiency. Techniques are devised measure efficiency at various speeds. The measurements incorporated into fitness function Ijspeert's original GA evolved. Interestingly, best evolved controller not only capable in similar manner real lamprey, but also with same (about 0.8). Moreover, it can exhibit wide range controllable speeds efficiencies.