Mitigation study and approach for safe drone operations in low-altitude environments

• Main Author: Sidharth, R Sai
• Other Authors: Low Kin Huat
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Mechanical engineering
• Online Access: https://hdl.handle.net/10356/158843
• Institution: Nanyang Technological University

With the recent increase in drone usage among organizations and hobbyists, there have been many accidents reported locally and overseas. Drones are built from commercial off-the-shelf (COTS) components as opposed to manned aircrafts. If these onboard components were to develop unmitigated faults during flight operations, it could lead to complete failure of the component leading to severe consequences such as fatality or damage to critical infrastructure. Thus, for further integration of drones into our day to day activities, safe mitigation techniques are necessary. This project presents a Fault Diagnosis Module (FDM) for actuator failures in quadcopters as a safe mitigation technique. To monitor the health of the UAV and provide early information on faults to take appropriate corrective action. A Hardware- in-the-loop (HITL) experimental setup is used to perform flight simulations. Partial and complete actuator fault simulations of varying level were carried out. The onboard sensor data was collected and subjected to preliminary analysis to understand the behavior and identify the controllability threshold. The FDM makes use of an online sequential fuzzy-extreme learning machine (OS-Fuzzy-ELM) algorithm which is trained using the fault behavior data at the controllability threshold to identify the location of the actuator fault based on the data provided. The average accuracy and macro-averaged F1 scores for the OS-Fuzzy- ELM fault diagnosis model are 80.2% and 78.0%.