Integration of Physical and AI Models Using Open and Interoperable Standards: A Model-Based Methodology for Autonomous Robot Development

Abstract

The development of Cyber-Physical Systems, particularly inthe field of autonomous mobile robotics, often relies onproprietary environments, which limit flexibility andinteroperability. This paper proposes a modular andopen-source methodology that enables the modeling andsimulation of such systems using non-proprietary tools. Themethodology integrates Functional Mock-up Interfacestandard for model exchange, Open Neural Network Exchangestandard for Convolutional Neuronal Networks algorithmsintegration, ROS2 as robotic middleware, and Gazebo as thesimulation environment. To validate the approach, weapplied it in the development of a mobile robot thatnavigates autonomously by following traffic signals. Thisimplementation demonstrates that these open technologiescan be effectively combined, overcoming common integrationbarriers among proprietary tools. The proposed workflowprovides a practical alternative to proprietary solutionsand demonstrates the feasibility of integrating openstandards for the development of autonomous robotic systems.