Biologically-inspired search algorithms for locating unseen odor sources

摘要

Many animals use air- or water-borne plumes of odor molecules to locate distant unseen resources. They offer excellent models for the development of robotic systems capable of orientation to chemical plumes. The best studied example of this behavior in biology is that of male moths tracking plumes of the female sex-attractant pheromone upwind to their source, a sexually receptive female. To more fully understand the complex interaction between the odor stimulus, sensory processing, interacting control systems, and ongoing centrally organized behavior, we have implemented a simulation organized around what is known about the sensory systems, behavior and control systems of real moths. The simulation environment is flexible and can reflect the stochastic nature of real environments. Within the biologically relevant parameter space, simple reflexive models are sometimes able to locate the odor source, but even the most successful models (comprising layered control systems and centrally generated behavior) fall far short of the performance of real moths. To try to understand why, we have employed a genetic algorithm to optimize the performance of the models. This approach has identified unique combinations of parameters that yield similar success rates, but display behaviors that look very different.