Occlusion-Aware AAV Trajectory Generation Method for Detecting Possessed Arms

Detecting dangerous objects using Autonomous Aerial Vehicles (AAVs) is crucial for security; however, challenges remain. Previous research has focused on global path planning, whereas local inspection trajectories that account for self-occlusion from an individual’s limbs have not been adequately addressed. The inspection target, humans, varies significantly in terms of body type and posture. Consequently, conventional simple circular trajectories often fail to capture self-occluded regions where dangerous objects are likely to be concealed, such as around the arms or waist, leading to inspection gaps. Therefore, this study proposes an occlusion-aware trajectory generation method that is adaptable to subjects with unknown shapes. This method improves the inspection coverage by up to 14 percentage points compared with the conventional circular trajectory and enhances the visibility of previously difficult-to-scan occluded areas. This study contributes to the realization of a future system capable of the high-precision detection of possessed arms.