Review of Recent Trends in Sensing Methodologies and AI Techniques for Non-Cooperative Aerial Target Recognition

The proliferation of non-cooperative aerial targets, particularly low altitude-slow-speed small RCS drones, presents a complex challenge for airspace security. Detection and identification of non-cooperative aerial objects is an intensive area of research. The aim of this article is to carry out a structured review of UAV/drones, fixed wing aircraft and other non-cooperative aerial targets’ detection and classification using machine learning (ML) algorithms. This article presents a systematic review of 184 recent studies (2019–2025) covering five key sensing modalities: Radar (RCS, Micro-Doppler, HRRP), Passive Sensing (5G/Wi-Fi/Radio Frequency (RF)), Acoustic sensing, Computer Vision and multi-modal sensing. Although individual sensing modalities have been extensively studied, existing reviews often lack a holistic operational assessment of multi-modal integration. Additionally, the review work introduces a novel operational suitability framework that maps each modality against critical deployment constraints, including detection range, Line of Sight (LOS) requirements, and environmental robustness. Furthermore, authors critically analyze the transition from classical statistical methods and outdated deep learning architectures to advanced deep learning architectures, specifically highlighting the emergence of vision transformers (ViT) and niche technologies such as integrated sensing and communications (ISAC). Finally, the review identifies persistent gaps in artificial intelligence (AI) based non-cooperative target recognition research and proposes a roadmap for future research in multi-modal machine learning (MML) and sensor fusion pipelines. Moreover, this review work would direct the research effort for further enhancing the aerospace, human safety, and important installations’ security.