Exploring possible organizations for visually-guided orientation

Abstract: This dissertation is concerned with exploring the behaviour and the internal properties of agent controllers which employ recurrent artificial neural networks, whose internal dynamics have been evolved by an evolutionary algorithm for coping with a visually-guided orientation task. Guided by a very simple fitness function, defined at behaviour level, the evolution allowed for emergence of different strategies, two of which are analysed in detail using traditional connectionist analysis methods. The examined strategies succeed in the orientation task by inherently handling a simple form of the object persistence problem in the limited environment, and by employing behaviours which include passive and active search, object tracking, obstacle avoidance, and a simple form of discrimination, each of which were observed to correspond to hidden unit activation subspaces.