Modeling obstacle avoidance behavior of flies using an adaptive autonomous agent

In the course of evolution flies have developed specialized visuomotor programs for tasks like compensating for course deviations, obstacle avoidance, and tracking, which are based on the analysis of visual motion information. In order to test models of the obstacle avoidance behavior in flies, we use computer-simulated agents that evolve parts of their sensor system and sensorimotor coupling with genetic algorithms. During a simulated evolution, these agents specialize a visuomotor program that enables the agents to avoid obstacles