Integrating Risk Assessment and Deep Learning-Based Anomaly Detection for Efficient Risk-Informed Decision Making for Swarm Flight Robot Systems

This study investigates the integration of risk assessment and deep learning-based anomaly detection (AD) for efficient risk management in the use of swarm drones. Deep-learning techniques and risk assessment methodologies were employed to develop a framework that supports timely decision-making and timely interventions, effectively mitigating potential adverse effects. The synergy between AD, risk assessment, and decision-making contributes to a robust and adaptive risk management strategy for complex systems. This study introduces a risk management index that consolidates individual drone risk assessments into a swarm-level risk index to facilitate immediate decision-making in critical scenarios. Practical examples were used to demonstrate the applicability of this methodology and its potential for enhancing safety, efficiency, and autonomy in swarm drone operations. However, the study results also highlight that the variability of the risk index, particularly when values fluctuate near the threshold, may limit its predictive reliability, indicating the need for further validation across diverse scenarios. The study results emphasize the significance of advancing risk management practices in the field of autonomous drone technology, fostering the leveraging of cutting-edge technologies for effective risk mitigation and decision-making in dynamic environments.