The global and regional need for air transport is coming back, as indicated by rising traffic and passenger numbers. This increase in demand is certainly something good for airports and the aviation industry, but it can also be a challenge. Over the past two years airports have lost quite a number of well-trained subject matter experts to other industries.
To compensate for this loss of expertise, airports can begin training new employees in specialist roles, but this will take time and money. Another option for airports is to invest in procedure changes enabled by new technologies, most notably: Artificial Intelligence (AI).
AI is probably best described as a collection of specialized subsets of related technologies. The simplest subset is AI itself. The AI function is programmed by a human to imitate human behavior.
A common aviation use case for AI is the automated planning and dispatch of fixed and mobile resources done by a Resource Management System (RMS). This AI function supports the operational efficiency of airport resource planning by suggesting fixed and mobile resource allocations automatically to the dispatchers based on a pre-defined rule set.
Another use case being implemented at many large hub airports is the routing function of an Advanced Surface Movement Guidance and Control System (A-SMGCS). Simplified, this routing function acts like a navigational system that provides routes to Air Traffic Controllers and, one day perhaps directly to pilots. The A-SMGCS routing function AI has also found its way into critical safety aviation functions, up and running already today.
The advantage of AI in the context of human-centered-automation, is that the defined rule-set can be explained to the operational users of the system. This helps them understand the output of the systems and enhance their trust in them. In case of failures, a human-in-the-loop can more easily take over with manual work.
A deeper subset of AI is Machine Learning (ML). ML uses statistical learning algorithms for constant improvement. An advantage of ML over AI is that the algorithms can improve over time. This can be very helpful in applications such as future Resource Management Systems, that will use an even bigger amount of flight schedule data, including historic information and forecasts, than today. Such systems will work more automized in regular operations and have the potential to unburden Asset Managers (e.g., for Gate and Stand) during their daily work. A continuous pattern matching in the background identifies residual irregularities, which then must be solved by human staff.
Deep learning can be described as trying to copy the function of the human brain by using neural networks. The application of deep learning in aviation and especially aviation safety functions is not yet common, due to the high cost for the needed computational power.
Seeing that AI is an option to improve the operational excellence of future airports and to reduce the current pressure of replacing specialized positions, the decision for or against AI is still a business decision. Creating a smart datahub is the first case to make, followed by use cases for data products on top of this. If the business case is compelling enough, procedure changes enabled by AI is a smart option to reduce costs. In the end however, every application of AI needs to be evaluated case by case, as airports operations vary from airport to airport. A growing number of mature use cases are around Computer-Vison and Video Analytics such as in the baggage area, people flow, FOD, etc.
Even if the business case for the implementation of AI is compelling and the intended AI function promises good operational improvement, there is still a point that should be taken into consideration in every AI decision: the techno ethical dimension. Tracking and improving the use of objects or improving resource allocation is a good way to use AI without any doubt. But when it comes to the tracking of humans, even if the data is anonymized, an additional complexity is added. From my point of view, every human should be informed that he or she is being tracked or monitored by a system and on the intention of the tracking. In the end, AI can bring many benefits, if used smartly and responsibly.
Edgar Ziller is T-Systems International’s Director Strategy & Partner Management Airports. Edgar joined the airport industry as IT project leader in 1997 and holds management positions within T-Systems for more than 20 years. Being an aviation enthusiast, he gained deep expertise in airport operations and progressive IT solutions. Reliability and trust, driving innovations and thinking ahead have always been Edgar’s firm principles.