Virtual Training System for the Autonomous Navigation of an Omnidirectional Traction Robot

Abstract

This paper presents the development of a virtual environment for the training of autonomous omnidirectional drive vehicle control. The virtual system considers the virtualization of structured and unstructured environments. Therefore, the virtual environment considers mathematical models of the omnidirectional robot in order to simulate more realistically the behavior and motion constraints of the robot. The integration of the control schemes is considered in the MatLab software, for which a communication between the Unity3D graphic engine and MatLab is considered through the use of DLL libraries. For the validation of the control algorithms on the virtual training environment, the construction of an omnidirectional traction vehicle with Mecanum configuration. In addition, the constructed prototype will be used for the identification and validation of the mathematical models that represent its behavior. Finally, a usability analysis of the developed training system is presented.