This paper presents how neural swimming controllers for a simulated lamprey
can be developed using evolutionary algorithms. A genetic algorithm is use
d for evolving the architecture of a connectionist model which determines t
he muscular activity of a simulated body in interaction with water. This wo
rk is inspired by the biological model developed by Ekeberg which reproduce
s the central pattern generator observed in the real lamprey (Ekeberg, 1993
). In evolving artificial controllers, we demonstrate that a genetic algori
thm can be an interesting design technique for neural controllers and that
there exist alternative solutions to the biological connectivity. A variety
of neural controllers are evolved which can produce the pattern of oscilla
tions necessary for swimming. These patterns can be modulated through the e
xternal excitation applied to the network in order to vary the speed and th
e direction of swimming. The best evolved controllers cover larger ranges o
f frequencies, phase lags and speeds of swimming than Ekeberg's model. We a
lso show that the same techniques for evolving artificial solutions can be
interesting tools for developing neurobiological models. In particular, bio
logically plausible controllers can be developed with ranges of oscillation
frequency much closer to those observed in the real lamprey than Ekeberg's
hand-crafted model.