Introduction
Various advanced technologies are being broadly introduced in different fields, and aviation is undoubtedly not an exception. Indeed, the implementation of automation in aviation may reduce the rate of accidents that sometimes result in hundreds of deaths, but it requires proper testing, training of pilots, and adequate certification procedure. For example, Boeing 737 Max plane crash in 2018 in Ethiopia caused the death of 189 people (Henderson et al., 2022). As the investigation showed, it was caused by insufficient training of pilots in working with the new artificial intelligence (AI) program, the Maneuvering Characteristics Augmentation System (MCAS) (Henderson et al., 2022). Therefore, the certification process was developed to ensure that pilots are capable of working with various artificial intelligence software, including MCAS. Modern jets are equipped with a so-called fly-by-wire automation system that connects pilots’ aircraft commands to the computer, which records and processes the input (Elias, 2019). Although these automation systems are primarily viewed as significant improvements to flight safety, insufficient training and lack of understanding among pilots led to unfortunate cases that could have been prevented by intense and continuous education.
Flaws of the Design Engineers
The investigation of the abovementioned Boeing 737 Max plane accident showed that the problem was caused by the error of MCAS; in other words, it was a faulty automation logic. In fact, it was found that the crash resulted from erroneous data collection by this AI system that led to the jet being excessively forced into a nose-down position (Crespo, 2019). Further research in this field revealed the correlation between system errors and the flight phase (Crespo, 2019). For instance, it was identified that about 80% of all faulty automation accidents occurred during the runway events like runaway excursions, incursion, and loss of flight control (Crespo, 2019). If these systems are improved and tested correctly, they “have dramatic potential to increase safety, reduce fuel and energy consumption, and increase accessibility to transportation services” (Donà et al., 2022, p. 1). However, the main problem nowadays is that the airworthiness certification process is still inefficient due to inadequate or incomplete flight test results (Liu et al., 2019). Although virtual flight test technology was developed in the United States to measure and evaluate various characteristics, it still requires appropriate and continuous adjustments.
Flaws of the DERS Certification Engineers
For the automation certification process to be productive and effective, it is crucial to have knowledgeable and experienced professionals assigned to the Designated Engineering Representative (DER) role. DER is a person who possesses technical knowledge and skills in the sphere of aviation engineering and whom a company assigns to conduct jet assessment according to the Federal Aviation Administration (FAA) guidelines (FAA, 2022b). Notably, every airline or private jet owner must apply for the FAA assessment process to receive flight approval (FAA, 2021). Moreover, DERs frequently employ data from simulation experiments that increase the accuracy of their assessment. However, simulation test results can only become reliable after they are very close to the actual test results (Liu et al., 2019). In addition to observing, evaluating, and testing, DERs must compile their findings in a documented form so that this technical report can be examined by the FAA committee (Liu et al., 2019). Still, approval from DERs does not guarantee that the problems with automation service and AI system will not appear during the flight. Hence, pilots should always be trained and prepared for such circumstances of the machine error.
The problem with DERs certification is that they must have vast knowledge and understanding of the engines they examine. Indeed, the certification for an aircraft is a lengthy and expensive process. For example, ground testing and test demonstrations may last over one year because experimentations should be conducted under various weather conditions (Mauery et al., 2021). It is essential for the system to be as precise as possible in poor visibility because many crashes occur during severe weather conditions. The 1973 Delta Airlines fatal crash in Boston happened due to heavy fog (Gawron, 2019a). Although this accident occurred before AI was introduced, it showed the importance of adequately calibrated and trained automation systems for jets. Additionally, aircraft manufacturers work in close collaboration with FAA and DERs to ensure that their jets meet the established requirements before they enter testing rounds (Mauery et al., 2021). Pilots, DERs, and FAA, should work in concert when checking automation systems because it is less the fault of one of the parties but rather the lack of communication and collaboration that results in failure.
Flaws of Continuing Airworthiness Certification Engineers (Aircraft Maintainers) at the Airlines Level
Another essential line of continuing airworthiness certification is a group of aircraft maintainers. The latter is a team of aviation engineers that validate if the jet is ready for exploitation by performing technical guidance, systems engineering, development, manufacturing, and testing (King, 2021). However, the main problem for a long time was that the certification process excessively focused on documentation (King, 2021). Most airworthiness certification maintainers believe that the introduction of model-based system engineering to the document-based approach improves the outcomes of testing and analysis (King, 2021). The test platforms that aircraft engineers utilize should include control laws, verification capabilities, a possibility for safety testing, pilots-in-loop simulation, troubleshooting abilities, and compiling flight test data (Yao & Liu, 2021). Since aircraft maintainers are only part of the large multidisciplinary FAA team, their task is to conduct appropriate safety testing and provide technical assessment reports to other specialists for evaluation (FAA, 2022a). Based on that information, FAA makes decisions about giving or denying the approval of flights for a specific aircraft (Civil Aviation Act, 2022). All of these procedures are necessary for the proper operation of jets and, most importantly, crew and passengers’ safety.
Apart from excessive bureaucracy, the problem of aircraft maintainers at the airline level is that they have relatively limited expertise. For example, a flight analyst focuses on flight information approval, while a propeller engineer can only approve propeller design, functioning, maintenance, and operation (Code of Federal Regulations, 2022). It appears that this issue stems from the educational level because engineers do not receive such training in colleges (Lopazan et al., 2021). Undoubtedly, it is better when every segment of an aircraft is checked and adequately examined. However, it is crucial to have experts that have a holistic understanding of how various elements contribute to the jet’s functioning because it improves troubleshooting and may prevent accidents more effectively. Fortunately, there are private organizations that provide assistance in preparing airlines and individual aircraft owners for the certification process, and S-Plane is one of them. This company conducts requirement analysis, aircraft configuration services, surveillance system consultation, and operation management assistance (S-Plane, 2022). Furthermore, S-plane provides help in terms of the selection of the right software for their jets.
Since technology has not yet reached the potential of manufacturing aircraft that automatically fix internal operational problems and prevent crashes due to unfavorable external conditions, FAA and its engineering group must approve aircraft’s safety. Although in the early days of aviation, aircraft maintenance was limited to fixing broken elements, it has become much broader now (IATA, 2022). In fact, today, jet reliability should be attained in design, production, and operation (IATA, 2022). As the name suggests, designed-in reliability ensures intrinsic reliability “through the aircraft design adopted solutions” (IATA, 2022, p. 5). Production reliability can be defined as the validity that results from an exemplary manufacturing process (IATA, 2022). Operational reliability can only be achieved during the actual operation of an aircraft; hence, testing and certification procedures are essential prior to the real flights (IATA, 2022). For the most part, the three types of reliability depend on FAA certification engineers as well as airlines’ technical maintenance crew, not on pilots. Nevertheless, pilot training, not only in terms of aircraft control but also interaction with the AI software, plays an essential role in ensuring passengers’ safety and preventing accidents.
Flaws of Pilot Training Related to Highly Automated Cockpit and Flight System Complexity
The bottom level of flight safety is the operation of aircraft by pilots because when the jet is in the air, the outcome primarily depends on pilots. Indeed, the safety of the airplane and the people on it depends on how well they are trained and understand the installed software. According to Swihart et al. (2011), “the majority of collision avoidance systems on fighter aircraft depend on the pilot taking action whenever a warning is issued,” which is valid for nonfighter jets, too (p. 4). Even though an automatic ground collision avoidance system is installed in modern airplanes, it is the pilots’ responsibility to control the situation (Swihart et al., 2011). Therefore, developing an efficient and rigorous training curriculum for these pilots is critical. The curriculum should contain not only theoretical knowledge as well as non-technical and technical skills but also be dynamic and interactive (Soo et al., 2021). Furthermore, in the past, training was primarily concentrated on procedural and psychomotor competencies, while now pilots are taught how to integrate AI systems in jet control.
Flaws and sometimes fatal failures that occur due to human error likely result from personal factors or insufficient understanding of the AI systems in the aircraft, especially if they are new ones. Some of these intrinsic factors include visual illusions due to poor weather conditions, false assumptions, personality issues, and fear (Gawron, 2019b). The automation problems include the difficult interface of the software, lack of adequate explanation about how the automation system works, and absence of feedback (Gawron, 2019b). It is crucial to help pilots learn to use the automation system and to ensure that they integrate it into jet control and view it as an inseparable element of an airplane (Soo et al., 2021). Additionally, the training curriculum should ensure proper comprehension of automation functionality, develop aircraft awareness, and, most importantly, allow practice automation use (Soo et al., 2021). Continuous training and knowledge updates can be helpful for pilots to improve their understanding of aviation automation and reduce the rate of fatal accidents.
Suggestions for Improving Aviation Automation Certification
Despite the abovementioned flaws that still exist and which are under the amelioration process, some improvements can already be introduced. For instance, the study Bleu-Laine et al. (2019) proposed the implementation of a Model-Based System Engineering approach, which helps transfer document-centric to model-centric methodology. It may resolve the issue of excessive bureaucracy in aviation certification. Moreover, pilot training should include more simulation testing under various conditions that involve AI not only to educate pilots but also re-check the automation system. Overall, if automation is adequately adapted to aircraft so that the error rates are significantly diminished, it can bring economic benefits for airlines as well as reduce stress and pressure on pilots.
Conclusion
In summary, aviation automation certification is a lengthy and complex but necessary process required for ensuring safety during flights. Although artificial intelligence software was installed in all modern jets, fatal crashes still happen today, especially in bad weather conditions when neither the pilot nor the system has an objective view of the surroundings. Additionally, flaws at the level of certification or verification and validation of aircraft sometimes occur due to inadequate education and excessive focus on documentation. Although Federal Aviation Administration controls the entire certification procedure, it still requires improvement. The latter can be attained by introducing a model-based system rather than a document-centric approach. Issues can also be caused by inadequate training of pilots in utilizing AI because sometimes they lack clarity and understanding of how the system works. Hence, their continuous education should incorporate better articulation of the automation system and more simulation practice under different weather conditions.
References
Bleu-Laine, M. H., Bendarkar, M. V., Xie, J., Briceno, S. I., & Mavris, D. N. (2019). A model-based system engineering approach to normal category airplane airworthiness certification. AIAA Aviation 2019 Forum, p. 3344.
Civil Aviation Act, Statutory Instrument No. 8. (2022). Web.
Code of Federal Regulations. (2022). Title 14. ECFR. Web.
Crespo, A. M. F. (2019). Less automation and full autonomy in aviation, dilemma or conundrum? In 2019 IEEE International Conference on Systems, Man and Cybernetics (SMC) (pp. 4245-4250). IEEE.
Donà, R., Ciuffo, B., Tsakalidis, A., Di Cesare, L., Sollima, C., Sangiorgi, M., & Galassi, M. C. (2022). Recent advancements in automated vehicle certification: how the experience from the nuclear sector contributed to making them a reality. Energies, 15(20), 1-17.
Elias, B. (2019). Cockpit automation, flight systems complexity, and aircraft certification: Background and issues for congress. Web.
Federal Aviation Administration. (2021). Airworthiness certification (AWC). Applicant registration guide and AWC application process overview. Web.
Federal Aviation Administration. (2022a). Aircraft certification, safety, and accountability act summary. Web.
Federal Aviation Administration. (2022b). Designated engineering representatives (DER). Web.
Gawron, V. (2019a). Automation in aviation – Accident analyses. Center for Advanced Aviation System Development: MITRE Technical Report MTR190013. The MITRE Corporation, 1–15.
Gawron, V. (2019b). Automation in aviation–Guidelines. The MITRE Corporation, 1–10.
Henderson, A., Harbour, S., & Cohen, K. (2022). Toward airworthiness certification for artificial intelligence (AI) in aerospace systems. In 2022 IEEE/AIAA 41st Digital Avionics Systems Conference (DASC) (pp. 1–10). IEEE.
IATA. (2022). From aircraft health monitoring to aircraft health management. Web.
King, J. C. (2021). Utilizing model-based systems engineering to identify safety critical functions in airworthiness certification [Master thesis, Air University]. AFIT Scholar.
Liu, X., Xiao, G., Wang, M., & Li, H. (2019). Research on airworthiness certification of civil aircraft based on digital virtual flight test technology. In 2019 IEEE/AIAA 38th Digital Avionics Systems Conference (DASC) (pp. 1-6). IEEE.
Lopazan, A., Cook, S. P., Lawson, K., & Greiner, G. (2021). Introducing aviation regulations and airworthiness certification into the aerospace engineering curriculum. The American Institute of Aeronautics and Astronautics, p. 2000. Web.
Mauery, T., Alonso, J., Cary, A., Lee, V., Malecki, R., Mavriplis, D., Medic, G., Schaefer, J., & Slotnick, J. (2021). A guide for aircraft certification by analysis (No. NASA/CR-20210015404). NASA. Web.
Soo, K. K. Y., Mavin, T. J., & Kikkawa, Y. (2021). Mastering automation: New airline pilots’ perspective. International Journal of Human-Computer Interaction, 37(7), 717-727. Web.
S-Plane. (2022). Aviation services. Web.
Swihart, D. E., Barfield, A. F., Griffin, E. M., Lehmann, R. C., Whitcomb, S. C., Flynn, B., Skoog, M. A., & Processor, K. E. (2011). Automatic ground collision avoidance system design, integration, & flight test. IEEE Aerospace and Electronic Systems Magazine, 26(5), 4–11.
Yao, Z., & Liu, L. (2021). Research on aircraft-level physical integration verification test technology. International Council of Aeronautical Sciences, 1-6.