Karel Kohout - Dissertation Thesis

My work was motivated by two main ideas that can be summarized in two sentences. The first key sentence is “anticipation is not matter of a single mechanism in a living organism it happens on many different levels”. The second pillar of my work is the fact that “anticipation happens even without voluntary control”. The combination of these two sentences is the most important point in my work and together forms my original approach to anticipation. I believe that this brings novelty for the anticipation theory. Naturally research of anticipation at the beginning of this decade was focused on the anticipatory principles bringing advances on the field itself. This allowed me to build on those, look at them from a higher perspective, and use not one, but several levels of anticipation in a creature design. In this work I will present my unified theory for anticipation and describe eight different types of anticipation ordered in a semi-hierarchical structure. I call this approach the 8-Factor Anticipation. This presents another original idea in this work and that is the agent architecture that has anticipation built in almost every function. I will describe these categories and demonstrate them in detail with examples and verify them in simulations. My 8-factor anticipatory behaviour approach is based on the current understanding of anticipation from both the artificial intelligence and biology point of view. Through this work I will first introduce the topic of anticipation and describe which scientific field it belongs to. In Chapter 2 I will describe the state of the art on field of anticipation in detail and mention works and theories that contributed to my approach. The parts important for my research are further probed in terms of algorithms and mechanisms which are described in the Chapter 3. This is followed by the core of my own work in Chapter 4 which describes my theory, the suggested improvements and used mechanisms. This theory is then verified and evaluated in experiments provided in Chapter 5. The whole work is then summarized in Chapter 6 along with future work, possible application areas and the weaknesses of my approach.