Study On The Behavior, Behavior Selection And Evolution Of Artificial Life Based On Sensor-Motor Intelligent Schema

Generally, natural life means the animal which live in the earth, has the feature of propagation, growth and evolution, such as human being, animal and plant. The artificial life not only exhibit the outside behavior character of natural life system, but also simulate, expand and extend its inside capability. The natural life can make use of the environment and oneself information to decide how to act, and can study and evolve when conflict occurs. Behavior of natural life has the feature of self-organized, self-adapted, self-studied, self-evolved, and so on. Only when the artificial life reacts to the outside stimuli and inside state as the natural life, it can be called the real life. This paper intents to explore the essence of natural life behavior, guided by the idea of"human simulated"and"bionics", using the research result of schema theory in cognitive science and evolutionary computing, does the research in behavior and its selection and evolution of artificial life based on sensor-motor intelligent schema(SMIS).This paper puts forward that use the SMIS as the intelligent basic element to describe the behavior, and takes the behavior selection schema to depict the behavior harmonizing and organizing of higher decision making behavior. Through the designing and denoting of schema assimilation and acclimation process, this paper realizes the design of study and evolution between the artificial life and the world. This paper simulates the implement process of human sensor-motor intelligent, shows the balancing between the behavior intelligence varies from low to high and the control precision changes from high to low, achieves the control in behavior selection and behavior evolution. In terms of exploring the life computing principle and building the artificial life system, the research in this paper has important academic meaning and application foreground for.The innovations of this paper are:â‘ Using the SMIS to describe the artificial life behavior that establishes the coding foundation for behavior and behavior selection under evolutional computing. Referencing to the schema theory in cognitive science and human simulated intelligent control theory, based on the analysis of animal and human sensor-motor intelligent, this paper introduces the schema theory and human simulated intelligent control theory into the artificial life behavior study. This paper constructs the behavior denoting method based on SMIS, and gives the detailed design means, which settles the coding foundation, in terms of behavior and behavior selection, for the quantitative study in self-studying and self-organizing and for the evolution in structure and parameter.â‘¡Combining the priority degree behavior selection and the production system, this paper builds the behavior selection schema for artificial life, which lays a foundation for self-organizing and self-studying of artificial life behavior selection. Based on the SMIS behavior denotation, integrating with the previous research on behavior selection structures during the fields of animal behaviors, robotics and artificial intelligent, referencing to the previous production on behavior selection based on priority degree, combining with the characteristic of production system, this paper puts forward the behavior selection schema for artificial life, in order to harmonize the task and select the behaviors.â‘¢By simulating the study and evolution process of biology, applying the improved evolution computing method, this paper firstly realizes the parameter self-studying and the structure self-evolving for SMIS of behavior and behavior selection, then resolves the design matter for SMIS assimilation, acclimation and balance adjustment.Base on the analysis of relation between the biology studying and evolving and the schema operation, with the behavior and behavior selection denotation based on SMIS, this paper resolves the design question of SMIS assimilation and acclimation. From the view point of evolution, combining the genetic algorithm with the genetic programming, based on the improved algorithm, this paper realizes the parameter self-studying and structure self-evolving of SMIS, and accomplishes the study and evolution simulation of natural life.â‘£Taking the double pendulum acrobatic robot as the experiment device, which proves the validity of SMIS theory for behavior, behavior selection and behavior evolution.The double pendulum is a kind of robot system which has the simple structure but difficult to control, it is the inherent unsteady system, and has the feature of high-rank, multi-variables, nonlinear, strong-coupled and under-actuated. Based on the previous study on double pendulum, apply this paper research method to build the many kinds of acrobatic behaviors for double pendulum acrobatic robot system. Under this system, based on the SMIS, this paper respectively introduce the design method of behavior, behavior selection and behavior evolution, and analyze the experiment results, in order to prove the validity of relative research in this paper.In conclusion, this paper dose some work as follows: firstly, produces a new method for artificial life behavior system designing; Secondly, resolves the key problems of behavior selection system in structure re-employ, transplant, and behavior study and evolution; Thirdly, improves the pervasive significance of artificial life structure based on SMIS; Finally, provides the wide foreground for simulation and actual application of complex behavior. The structure proposed in this paper has strong potential in engineering application, and in favor of constructing and studying various artificial life systems.