Research On The Robot’s Learning Algorithm Based On Artificial Emotion

With the development of computer and artificial intelligence technology, the use ofrobots has expanded from the traditional industrial manufacturing field to the military,civil, marine exploration, the moon and Mars detection, etc. These application fields areno longer simple structured environment, but also some extreme or complex environment.In this case, we need the robot to have higher performance, as the current level of theintelligent robot has failed to meet the needs of complex environment. The traditionalrobot control mechanism are based on the cognitive way, this control mechanism can’tmeet the demand of extreme environment or the complex environment, so we shouldimprove the robot-study way from control mechanism. While the research aboutemotional intelligence has made some achievements, so joining the artificial emotion intothe intelligence control of robot has become a new research interest.In this paper, the research was carried out on the robot learning algorithm, we addedthe factor of the emotion into the robot learning algorithms to build the robot learningframework based on artificial emotion. Then we used it for CMAC network learningalgorithm, genetic algorithm and reinforcement learning algorithm respectively toimprove robot’s comprehensive performance.Firstly, CMAC network learning algorithm, genetic algorithm and reinforcementlearning algorithm were analyzed, and combined with the process of human brainprocessing information, the idea that putting the emotion factor into the process of robotlearning was put forward.Then the robot learning framework based on artificial emotionwas built, which was the foundation to improve CMAC network learning algorithm,genetic algorithm and reinforcement learning algorithm respectively.Secondly, the characteristics of the natural emotion were researched deeply. Thenthe emotion model based on Euclidean Space and the emotion model based on cognitivereasoning were established, and the simulation results verified the validity of two models,proving that artificial emotion motion could be used in intelligent control of the robot.Last, on this basis the emotion model based on fuzzy reasoning was established, whichwas applied in the following simulation experiments. Finally, in the Matlab environment, the robot survival experiment and robotnavigation experiment were set respectively to test three improved algorithm.Experimental results proved the design’s rationality of the robot learning frameworkbased on the artificial emotion and the feasibility of improved algorithm to improve theoverall performance.