The role of digital twins in the certification of the Advanced Air Mobility (AAM) systems

Abstract

AAM (Advanced Air Mobility) is an emerging field in aviation that concentrates on developing AAM/electric vertical takeoff and landing (eVTOL) aircraft for urban air mobility. AAM can use human-centered design to provide an advanced emerging environment of the eVTOL aircraft and its operating environment, allowing for a more efficient and cost-effective development process – concentrating on human factors/ergonomics, training, certification, and qualification. According to the EASA Artificial Intelligence Roadmap (EASA, 2023a), industrial actors expect the first crew assistance/augmentation developments in 2022–2025. Automation will gradually ramp up to human/machine collaboration between 2025 and 2030, culminating with human supervision or fully autonomous systems the year 2035 after. To achieve those milestones, the EASA guidance for Machine Learning proposal (EASA, 2023b) for Level 1 Artificial Intelligence (assisting humans) and Level 2 Artificial Intelligence (human-machine collaboration) aims to proactively address forthcoming EASA guidelines and safety standards about machine learning (ML) applications with safety implications. It guides applicants who incorporate AI/ML technologies into systems designed for safety or environmental purposes.Moreover, it provides guidelines covering the following building blocks that lead to Trustworthy AI: AI Trustworthiness Analysis, AI Assurance, Human Factors for AI, and AI Safety Risk Mitigation. CAE - Purdue proposes a research case study for digital twins in AAM that targets designing and remote testing prototypes - eVTOL aircraft simulator devices. By creating a digital twin of the AAM flying/simulator device, designers (Purdue Human Factors team – CAE network) and Subject Matter Experts (SMEs) aim to test different configurations and scenarios. This allows the research team to identify human factor – certification challenges before building the physical prototype. The Artificial Intelligence (AI) research roadmap of the Purdue School of Aviation and Transportation Technology (SATT) focuses on the potential to increase the effectiveness and efficiency of AAM design by providing a realistic and immersive experience (lean process for training/certification, transition to A.I. - AAM environment). In addition, this study concentrates on mitigating residual risk in the 'AI black box.' The Artificial Intelligence (AI) certification outcomes and learning assurance challenges were analyzed and evaluated.