Data-Driven Identification Method and Simulation Modeling of a Ground Robot

Abstract

Autonomous mobile robots rely on the environment features to operate. In a framework in which agents operate autonomously from each other and only use their on-board sensors and computing power, an accurate simulation environment has to be set up. Accuracy is increased by producing a good model of the robot agent itself. In this paper, a data-driven identification method is exploited to design a black-box model for numerical simulations. A MATLAB/Simulink-ROS-Unity3D hybrid environment is considered as simulation scenario, to be easily connected to the on-board real hardware.