Uncovering drone intentions using control physics informed machine learning

• DOI: 10.1038/s44172-024-00179-3
• Publication Date: 2024-02-24T14:02:04Z
• Authors: Adolfo Perrusquía; Weisi Guo; Benjamin Fraser; Zhuangkun Wei
• Keywords: Drone; Unobservable; Proxy (statistics)

Abstract Unmanned Autonomous Vehicle (UAV) or drones are increasingly used across diverse application areas. Uncooperative do not announce their identity/flight plans and can pose a potential risk to critical infrastructures. Understanding drone’s intention is important assigning executing countermeasures. Intentions often intangible unobservable, variety of tangible classes inferred as proxy. However, inference drone using observational data alone inherently unreliable due learning bias. Here, we developed control-physics informed machine (CPhy-ML) that robustly infer classes. The CPhy-ML couples the representation power deep with conservation laws aerospace models reduce bias instability. achieves 48.28% performance improvement over traditional trajectory prediction methods. reward results outperforms conventional inverse reinforcement approaches, decreasing root mean squared spectral norm error from 3.3747 0.3229.