The Cognitive Development Learning Model and Its Application in Autonomous Navigation of Mobile Robot
Revista de la Facultad de Ingeniería · 2016
In terms of the navigation problem of the mobile robot occurring in unknown environment, based on the cognitive and developmental mechanism of the biology, a cognitive development model for the autonomous learning navigation of the mobile robot is designed. Through autonomous insertion and deletion of neuron nodes in the neural network, the neural network with a dynamically adjusted structure is designed and the biological characteristics of living beings are imitated, thereby obtaining a network scale that matches the application requirements. The thermodynamic process is used to simulate the asymptotic properties of the animals. The cognitive learning algorithm is designed from the Monte Carlo method, the Metropolis algorithm and the annealing algorithm, and it is theoretically proved convergent. Based on the formation of the cognitive development process, the autonomous navigation activities of the mobile robot is under study. The experimental results show that, with the help of the cognitive development model, the robot automatically acquires knowledge and accumulates experience from the environment through cognition and development like animals. And the robot gradually forms, develops and improves the autonomous navigation skills in a self-organized way.