Robot Imitation Learning Based on Gaussian Processes

To acquire the control strategy in robot imitation learning, a Gaussian processes ( GP ) regression model based on GP is used to construct the mapping relationship between actions of teaching robot and perception by the sample data of the teaching behavior. Then, the mapping relationship is used as control the strategy to imitate the teaching action. The Braitenberg vehicles are used as simulation subject to study phototaxis imitation learning action. Simulation results show that robot imitation learning based on Gaussian processes is effective. Furthermore, the simulation results in various task environments indicate that the method is adaptive.