Digital Twin for Firmware and Artificial Intelligence prototyping

Abstract: The forth industrial revolution has risen the born of new mega trends for the improvement of the time to market and the spare of resources in the development and manufacturing of a new product. Among these trends, the Digital Twin (DT) is the one of major interests for developers and strategy analysts. The perfect transposition of a real entity into a digital environment enables the exploration and testing of the different components within the defined object, taking a further step towards a perfect correct-by-design approach. STMicroelectronics (ST) is exploring the benefits that this technology offers to the developers. The company’s primary focus revolves around the creation of SystemC models for the manufactured components so that a co-simulation between an Hardware (HW)/Software (SW) platform and a kinematic simulator is possible. This innovative approach facilitate the comprehensive validation of the designed Firmware (FW), relying on the intricate interplay with sensory aspects influenced by both device behavior and environmental circumstances. Furthermore, many applications nowadays implement an Artificial Intelligence (AI) algorithm: its performance is strictly dependent on the quality of the signals sensed and on the dataset on which the model is built. The creation of a proper DT allows to implement its development during the design phase, creating not only a valid AI for the real product, but also improving the quality and the performance of the model built. This conclusion is proven through the construction of a simple robotic arm implementing an anomaly detection algorithm based on a Machine Learning (ML) model.