SAFETERM uses computer vision techniques to enhance flight termination systems & procedures for Medium-Altitude Long-Endurance RPASs
AI-GNCAir studies the takeup of artificial intelligence in guidance, navigation & control for aerial applications
Over the last decade artificial intelligence (AI) buffs have developed new algorithms and learning strategies that have ushered in what has come to be dubbed the fourth industrial revolution. AI and machine learning (ML) take up is therefore now booming in many sectors, and the aeronautics sector, thanks to the technology multinational GMV’s expertise, is no exception.
In this overall context the European Defence Agency (EDA) has awarded GMV two projects: SAFETERM and AI-GNCAir, two of the most advanced initiatives being carried out by GMV in this field.
SAFETERM’s purpose is to improve the flight termination procedures of Medium Altitude Long Endurance (MALE) RPASs.
Under this overarching goal its main remit is to increase the general safety level in emergency situations involving multiple failures including C2 datalink loss, i.e. without remote pilot intervention. In this situation it would switch to safe, alternative landing areas by means of computer vision (CV). This involves an extremely complex task that represents a huge advance on traditional image-processing techniques.
Amongst all the possible CV applications, SAFETERM is based on area recognition: what areas does the image show and where are they? In this project, another of EDA’s goals is to weigh up the challenges of using AI in aviation. Can AI powered functionality be developed for reala avionics hardware and software? This leads to another important aspect of the project related to the certification and standardization support activities. GMV is currently a member of EUROCAE WG114 – SAE G34: a joint standardization initiative to support the artificial intelligence revolution in aeronautics, especially in safety critical systems.
Artificial Intelligence in Guidance, Navigation and Control for Aerial Applications (AI-GNCAIR) studies the takeup of artificial intelligence in guidance, navigation and control for aerial applications.
Led by GMV and carried out in collaboration with the Telecommunications and Information Processing Research Center of the Polytechnic University of Madrid (Centro de Investigación en Procesado de la Información y Telecomunicaciones de la Universidad Politécnica de Madrid: UPM-IPTC), AI-GNCAIR sets out to recommend a generic GNC architecture for the safe use of AI-based algorithms in the aeronautics sector. The second phase will involve a practical-case simulation to compare the new algorithms’ performance against that of traditional data fusion techniques.
AI-GNCAIR forms part of EDA’s strategic research agenda under CapTech GNC, which studies how to integrate AI into GNC systems and the necessary roadmaps to close the EU’s technology gaps.
Some of the standout features of data management and sensor readings for navigation tasks are: security, preventing any tampering with the readings: integrity, to ensure and monitor calculation-flow data; and availability, to ensure data flows are never cut off. AI algorithms have to pick up any signal interference, incorrect sensor readings and even forecast any missing data due to the above circumstances.
Some of the fields AI-GNCAir is focusing on are robust data acquisition, efficient data-fusion protocols, data-fusion computation complexity management and dynamic sensor selection to ensure unbroken availability.
AI and ML are generic terms embracing a huge variety of data-optimization, control and processing techniques applicable to practically any sector or system. Aerial vehicles could benefit from this advanced technology, guaranteeing greater autonomy and safety and allowing human operators to input higher-level information and exert greater supervision.