| Aiming at the self-balancing control for two wheel self balancing robot,in this paper,based on the principle of MDB cognitive development mechanism which draws on the dual system theory of human brain in cognitive science,the development algorithm model was established with different computing structures.These models allow the robot to learn the self-balancing control rules by interacting with the environment.Aiming at learning problem for MDB,based on the evolutionary strategy algorithm(ES),a learning automate model(ES-CALA)was established,which is a optimization algorithm,and it consists of three parts: action function,output and learning function.The experiment results show that ES-CALA has the feature of autonomous learning.Aiming at the self-balancing control,multiple ES-CALA was combined together to form a CALA combination,and the cognitive development model(CDM)of if-then structure was established based on the combination.The model consists of a motivational model,a control loop and a learning loop.Its function was introduced in detail,which was used in self-balancing control.Autonomous learning ability and adaptive experiment were established,the results showed that the robot can learn control rules by itself,and it adapts to changeable environment,showing the feature of MDB.When the control rules were used to control the inclination angle,there was a small angle fluctuation.For the fluctuation,a CDM having with the ability of autonomous learning based on transfer function was established,and it could eliminate the above small angle fluctuation.The model consists of control loop,controller evolution loop and model evolution loop.Under Simulink/MATLAB platform,the CDM effect was verified.In this paper,a development model was established based on ES-CALA,and an improved method was proposed.for the ES-CALA,Cascaded networks can be built.For world models,neural networks,support vector machines,or random forests can be used to solve more complicated problems. |
PDF Full Text Request