A Study of Software Health Management in Unmanned Aerial Vehicle Systems

Abstract

Unmanned Aerial Vehicles (UAVs) must operate their installed software without errors to complete their mission. Software faults cause errors in the software. Most of them are removed through the Verification and Validation process in the development phase. However, developers and testers cannot eliminate all faults due to the inability to account for all uncertain environments and conditions. System health management is the technology that helps UAVs successfully perform missions by mitigating errors during operation. However, there are errors that the existing UAV System Health Management cannot diagnose because errors are not deterministic nor explicit. This paper analyzes two UAV systems, PX4 and Ardupilot, to categorize software errors. It also assesses whether the existing UAV System Health Management can diagnose these errors. Our study can help developers and researchers better understand the characteristics of software errors in UAV systems and set the direction for future work in UAV software health management. © 2023 Copyright for this paper by its authors.