Strengthening Data’s Role in Aviation Training

This is the second in a series of articles on data’s expanding use for commercial aviation training.

While training organizations use this resource to help establish individual performance trends and other outcomes to ensure the efficiency and safety of pilots in ground-based and in-the-air training events, some stakeholders are using databases to advance evidence based training and competency based training and assessment – and much more.

In this article we’ll also read about efforts by suppliers of software and other data-related products and services to bring to bear an eclectic mix of technologies to permit this resource to be used through a pilot’s career learning continuum. While cloud computing and enhanced security are representative enablers of data-based systems, simulation and training companies, and their aviation customers are stepping up their activities to integrate AI and other underpinnings in their learning programs.

Two High-Level Insights

Aylin Özgeneci, Simorg Product Manager, and Devran Gözen, Chairman of Gözen Digital Aviation (Gözen Digital), provided an initial, high-level insight on this sector’s use of this resource. The executives pointed out they are partnering with independent training centers and airlines which increasingly seek to track individual performance and identify airline-wide training trends that lead the company to generate comprehensive trend analysis reports. “Our partners require detailed insights into instructor performance trends,” they noted, and said they also supported a customer who shifted from competence-based to evidence-based training methods. “We facilitated detailed monitoring and reporting capabilities by embedding evaluation methodologies into their system that constitute various types of data into a meaningful whole.”

Another place-holder in this discussion validates and quantifies the vast amounts of data being generated across organizations for training purposes. Andy O’Shea, CEO of The Airline Pilot Club, and former Head of Training for Ryanair, said one thing that has become apparent to regulators and other community members, including those at APC, is the amount of data that we’re talking about that is actually generated in training. For instance, during a 1.5-hour simulator training event, “You’re going to generate 100 gigabytes of data without any problem. The flight simulation guys will generate 200 gigs to one terabyte [TB] of data on the same amount flying on the two-hour session.” Of significance – a TB equals one trillion bytes.

Beyond these overarching perspectives, a tech deep dive is in order to provide insights on how several representative, fielded products and one in development permit training organizations to reap returns on investment from data-based systems.

Product Profiles

One of Gözen Digital’s flagship products, Simorg, is designed to integrate various aspects of training management, ensuring data security and customization. Gözen and Özgeneci also hit upon another important notion – of data being gathered from multiple sources for use. In this instance, Gözen Digital’s systems combine multiple data sources, including flight operations, simulator performance, and maintenance reports, to create comprehensive training programs and utilize advanced algorithms to analyze data, identifying trends, hazards, and areas for improvement. “Simorg enables partners to manage and utilize their data efficiently for training purposes, safeguarding the uniqueness and confidentiality of each client's data,” they emphasized.

The two executives also called attention to the significance of their business model – combining deep aviation knowledge with advanced digital solutions. “Our portfolio stands out due to this holistic approach to training management. Developed from the practical needs of Gözen Group's own simulator center - IFTC, Simorg addresses every step of the training process, from initial planning to billing and compliance.” They further explained the system is modular, with each part triggering and supporting the others, allowing for seamless data integration and utilization. “This holistic design, combined with our extensive industry experience, sets Simorg apart from other solutions in the market​​. Our partners who have global presence operating under different regulations, rules, and operational practices benefit from our capability to interpret and refine diverse datasets, enriched by our industry knowledge.”

This data includes, but is not limited to: engine thrust settings; airspeeds at critical points of the flight: take off, cruise, approach and go around; pitch of the aircraft at critical points – take off / landing; landing depth on the active runway; tail clearance during take off and landing; and SOP compliance.

PilotApp also provides other data-related technology enhancements. Jesse said that in another instance, by providing FDM/FOQA data directly to the individual Pilot, latency in the safety system is removed. He added, “Typically FDM/FOQA data is analyzed by a safety team and stats fed back, where appropriate, weeks or months later. PilotApp provides personalized post-flight feedback to pilots within minutes and their statistical performance is refreshed daily – allowing pilots to understand their performance immediately rather than waiting for audits, line checks and simulator evaluations.” Further, L3Harris’s existing customer base has seen a significant decrease in event rates when providing personalized feedback back to their pilots.

Jesse walked CAT through one use case for PilotApp by selecting a representative training event, a FFS scenario. The product manager noted in the CBTA/EBT domains the key issue is creating tailored training solutions for the pilot. “An annual recurrent check requires a two-day event: Day 1 is evaluation with Day 2 the competence development event based on the Day 1 evaluation. PilotApp has a shared statistics capability allowing pilots to volunteer their on-the-line performance with their instructor at our London Training Centre ahead of the recurrent training session. This only enhances training delivered through LOFT [line-oriented flight training] exercises by personalizing the training events to that individual’s performance data. For example, it is valuable in cases such as reducing the risk of tail strikes.”

AI to Strengthen and Expand Data Usage

The commercial airline industry is using AI with increasing frequency across its enterprise –to enhance predictive maintenance, gain efficiencies in flight operations and even make happier and more satisfied passengers starting with the reservation process – but let’s not forget training.

APC is pushing the envelope on using AI for training with its Amelia.2 AI-powered platform. The platform essentially ingests and analyzes structured and unstructured data from pilot selection and training processes to provide insights, identify areas for improvement and generate personalized training plans. Cedric Paillard APC’s COO and architect of Amelia.2, and Andy O’Shea, who concurrently spoke with CAT, both emphasized “the platform is designed to be a tool to assist human decision-makers, not replace them.”

The commercial airline industry is using AI with increasing frequency across its enterprise – to enhance flight operations (above, left) and other purposes – but let’s not forget training (above, right). Source: Alaska Airlines; L3Harris Commercial Aviation

In early proof-of-concept use cases, the platform is assisting with pilot recruitment, training and personalized learning. Specifically, APC brought Amelia.2 to bear with Avianca, where the platform was used to screen and assess 1,776 pilot candidates, narrowing the pool of aspiring aviators to 180 in two months. Paillard pointed out Amelia.2 was able to identify key competency gaps and generate personalized training plans to address them, improving the success rate of candidates moving forward.

Of further interest and significance to the broader simulation and training community in and beyond aviation training, are Amelia.2’s technology building blocks – increasingly visible and common in adjacent high-risk training enterprises, be it for defense or safety critical industries. Two of the technologies include Large Language Models (LLMs) and a SaaS (software-as-a-service) cloud computing model. Paillard initially commented, “The reason why LLM is used is because you can actually take unstructured data, including handwritten notes, and you can actually use it.” And the SaaS cloud computing model – to best support linking to other systems to allow the data to be entered and available to designated airlines, ATOs and other stakeholders.

APC’s O’Shea concluded, “As the platform continues to evolve with upcoming releases like Software 3.0 which will further integrate advanced features like LC/NC (Low-Code/No-Code) solutions and autonomous agents, APC is poised to lead the industry into a new era of personalized, data-driven training methodologies.

L3Harris is also using AI to personalize pilot training activities and optimize the pilot training journey. Jesse explained, “It’s an effective recommendation engine to identify appropriate training courses in response to instructor identified route causes. AI is also being used to identify emerging trends that allow pilots and instructors to take proactive training steps before events are triggered, which, in turn, allows operators to update LOFT exercises to minimize operational drift.” The product manager concluded that with the use of AI evolving at a rapid pace, “the Data Analytics team are continually reviewing the latest technology to see how it can be adapted to benefit our industry and fit the company mission of innovating to create safer skies.”

Currently, there are no AI or XR-based training data integrations within Gözen Digital's portfolio. However, the executives added, “We continually work on integrating emerging technologies into our solutions to enhance our offerings and meet evolving industry needs.”

Down-stream Developments

The aviation training sector is also increasingly recognizing the importance of real-time data analysis and AI-driven insights triggered by big data for forecasting. Gözen and Özgeneci pointed out training centers are expanding internationally, and this requires standardized data collection and consolidated reporting across different locations and methodologies. “This trend highlights the need for integrated management systems like our solution-Simorg, which can handle diverse data types and provide unified, actionable insights. This dual approach ensures that data is not just collected but also analyzed in a way that supports informed decision-making and enhances operational efficiency. We anticipate a growing demand for such comprehensive solutions to manage and optimize training processes effectively.” Additionally, team Gözen expects an increase in the use of cloud computing to manage and store large volumes of training data, providing scalable and flexible solutions for training organizations.

APC’s go-to-market plan for Amelia.2 has two major waypoints through the end of 2024. There will be a program status announcement concurrent with APATS 2024 and Amelia.2 will be initially shown to the community at EATS 2024. As these milestones approach, APC remains committed to pushing the boundaries of what AI can achieve in aviation training, setting the stage for what could be transformative in the industry.