The ability to make decisions autonomously and act on them is a core theme of artificial intelligence. Nevertheless, transport drones today are not able to detect and correct errors in their system autonomously. This leads to drone crashes and can result in high property damage and personal injury. To ensure the functional safety of transport drones, the associated partner Viafly provides Fraunhofer IOSB-INA with data on error-free and faulty drone flights. Fraunhofer IOSB-INA uses its know-how in the field of automatic error detection and handling to analyze the data. The aim is to detect errors at an early stage and initiate countermeasures in time.
Transport drones are used in extreme environments and are therefore subject to environmental influences such as temperature fluctuations, crosswinds and the like, which affect the drone’s dynamics. Erratic flights and even crashes are the result of these disturbances. In recent years, various methods for automatic reconfiguration of transport drones have been presented. However, the main focus of these methods is on the reconfiguration of numerical control variables. So far, these innovative methods have hardly been tested in practice. Modern drone systems are usually based on a control system that cannot react to unforeseen disturbances such as winds, etc.
The goal of this project is to use methods of model-based failure analysis and reconfiguration, a subfield of artificial intelligence, to detect the effects of these disturbances on the transport drone and determine appropriate countermeasures to stabilize the drone’s flight and avoid crashes. For example, a failing rotor of the drone should be intercepted by adjusting the motor outputs to the other rotors.
Viafly GmbH provides recorded data of the flight controls including position data as well as motor and battery data. These are unprocessed raw data. Within the scope of the project, these data will be processed and analyzed so that they can be used directly in follow-up projects. Furthermore, the potential regarding fault identification and automatic reconfiguration, which is contained in the data, will be revealed; in follow-up projects, this potential will be used to develop innovative solutions.
Fraunhofer IOSB-INA can further expand its competencies in the field of artificial intelligence for industrial applications, especially the steadily growing market of drones. This experience will be transferred to companies, especially SMEs, in the future. In addition, Fraunhofer IOSB-INA can further enhance its scientific reputation by publishing the results of the evaluation.
01.11.2019 – 30.04.2020
Dr. Alexander Maier (firstname.lastname@example.org), Fraunhofer IOSB-INA
Dr. Aaron Gryzia (email@example.com), Viafly GmbH