Study on chaotic prediction of adaptive behavior for wheeled mobile robot

Abstract

Based on a group of one-dimensional time series of adaptive behavior obtained via Artificial Intelligence (AI) theory, this paper analyzes the evolution of adaptive behavior by chaos and phase space reconstruction theory. The autonomous prediction model is constructed by Artifitial Neural Network (ANN). The phase space of one-dimensional time series of adaptive behavior is constructed based on phase space reconstruction theory. The values of embedding dimension and the largest Lyapunov exponent of the phase space can verify that the system is chaotic. The adaptive behavior prediction model is constructed by the embedding dimension and neural network theory. The simulation results show that the prediction model can forecast the adaptive behavior exactly in one step. The robot can forecast the adaptive behaviors autonomously based on the real-time information. This method provides a theoretical basis for the enhancement of the robot adaptive behavior forecast accuracy in nonlinear and non-structural environments.