Research Of Cognitive Model And Behavioral Control Of Artificial Fish In Virtual Marine Environment

Research on artificial life (ab. ALife) is rising now. It is the outcome of the two most energetic, promising and fascinating leading disciplines in 21st century, that is, life science and information science. Life phenomena as the research object, ALife mainly studies the mechanism and engineering implement technology of life procedures and aims at expanding life performance of human beings. It plays a very important role in current science.ALife is a cross-discipline research area and its content is increasingly enriched in the past twenty years. Animat, in other words, artificial animal, which simulates various creatures in nature by means of computer, has become a major branch of ALife. Artificial fish is one kind of animats. It displays multiple life properties and performance based on simulation of modality, habits and behaviors of natural fish, builds the virtual marine environment for the fish and provides people with a vivid virtual marine world. Artificial fish is a good research platform of many fields such as animal behaviors, computer graphics, computer simulations, virtual reality, etc, and it can be applied into biologic and marine teaching demonstration, marine world visualization, education and entertainment. It has great values both in research and application.This dissertation takes artificial fish as the research object and studies its cognitive model and behavioral control methods through combination of artificial intelligence, artificial life and virtual reality. The main contents and innovations are summarized as follows.1. Architecture of artificial fish including perceptive, cognitive and behavioral sub-systems is built on the basis of hybrid-agent principle. Functional design of every sub-system is given and inner state model of artificial fish is proposed.2. A memory-based cognitive model of artificial fish-MBCM is put forward and built according to the biological principle and memory mechanism of natural fish. Structure and working principle of the cognitive model are given. Five levels of autonomous character modeling are explored and their respective features are analyzed. Method of information coding, storage and access during memory process are designed and memory-based cognitive ability of artificial fish is realized.3. A dual behavioral control structure of artificial fish composed of high-level behavioral planning and locomotion model is proposed based on integration of artificial intelligence and artificial life. Different strategies of fish behaviors such as prey, escaping, school and so on are realized through high-level behavioral planning algorithm which breaks through the limit of single behavior mode of artificial fish. The relationship between locomotion velocity and inner state including stamina and hunger is given. Life properties and intelligent behaviors of natural fish are simulated more reasonably and efficiently.4. A group behavior cooperative control method of artificial fish based on multi-agent is proposed, expanding individual agent modeling into group agents modeling. On the basis of message center, real time perception of artificial fish to dynamic environment and other agents is realized through message distribution, reception and disposal mechanism. Then current intentions are carried out using high-level behavioral planning and locomotion model. Therefore, cooperative control of artificial fish group behavior is realized.Based on the above theories and methods, a virtual undersea world simulation system is designed. The system will use real-time rendering method in virtual reality to model the virtual marine environment and display the special effects, providing the artificial fish with a vivid habitat. Models and algorithms proposed in this dissertation are applied into design of artificial fish and make them intelligent characters with autonomous behaviors and life properties.