Distributed learning with biogeography-based optimization: Markov modeling and robot control

Biogeography-based optimization (BBO) is an evolutionary algorithm that is motivated by biogeography, which is the science that describes how biological species are geographically distributed. We extend the standard BBO algorithm to distributed learning, which does not require centralized coordination of the population. We call this new algorithm distributed BBO (DBBO). We derive a Markov model for DBBO, which provides an exact mathematical model of the DBBO population in the limit as the generation number approaches infinity. We use standard benchmark functions to compare BBO and DBBO with several other evolutionary optimization algorithms, and we show that BBO and DBBO give competitive results, especially for multimodal problems. Benchmark results show that DBBO performance is almost identical to BBO. We also demonstrate DBBO on a real-world application, which is the optimization of robot control algorithms, using both simulated and experimental mobile robots.