Electrical Engineering-Systems Dept.

*** SEMINAR ***

Shlomo Israel

(M.Sc. student under the supervision of Dr. Amiram Moshaiov and Prof. Dana Ron)

on the subject:

 Evolution Bootstrapping using Multi-Objective Optimization

Aggregate fitness selection of individuals is known to suffer from the bootstrap problem, which is often viewed as the main inhibitor of the widespread application of aggregate fitness selection in evolutionary robotics. There remains a need to identify methods that overcome it, while requiring the minimum amount of a priori task knowledge from the designer.

To resolve the bootstrap problem we suggest a novel two-phase method and apply it to the evolution of robot controllers. In the first phase, the method exploits multi objective optimization to develop a population of controllers that exhibit several desirable behaviors. In the second phase, aggregate selection of individuals is carried out using the previously obtained population as a seed population. The multi objective phase is based on adaptation of the well-known NSGA-II algorithm for coevolution. The proposed bootstrapping approach is demonstrated using simulated coevolution of robotic soccer players. The proposed approach is compared with a conventional one, which has been suggested by other researchers. Assessment is done by two non-traditional comparison procedures, including one that is introduced in this thesis for the first time.  The results demonstrate the potential advantage of the suggested two-phase approach over a conventional one.