Meeting Abstract

P3.59  Jan. 6  Testing the Adaptive Value of Vertebrate Tail Stiffness, II: Computer Simulation CUMMINS, M.A.*; ROOT, R.G.; LONG, J.H.; LIEW, C.W.; Lafayette College; Lafayette College; Vassar College; Lafayette College liew@cs.lafayette.edu

To test the hypothesis that tail stiffness was a locomotor adaptation in early vertebrates, we use genetic algorithms to evolve a virtual fish-like swimmer. The virtual swimmer is a fully dynamic, feed-forward model of our robotic sea squirt tadpole larvae (�Tadro�), and, after validation, provides a way to test many more evolutionary scenarios than is possible under the constraints of robotic simulation. Specifically, we focus on the effect of vertebrae on the bulk modulus and spring stiffness of the tail. Virtual fish of variable tail stiffness and length compete and breed through an iterated genetic algorithm suited for a high-dimensional search space (approximately 8 dimensions). Fitness is measured as a minimization of time in a two-dimensional phototactic approach to a stationary suspended light source, in addition to the ability to hold station around the source. The results from the virtual fish are compared with those obtained from the Tadros. Supported by the National Science Foundation grant DBI-0442269