This article is part of a series involving the participants of the AI program by Leonard. The program has been specifically designed to accelerate the adoption of IA technologies within the group VINCI. It consists in a five-month incubation period where selected VINCI collaborators follow a learn by doing process where they codevelop an AI-based use case under the coaching and mentoring of the Leonard Team and Eleven consultants.
Machine learning models have recently begun to challenge traditional ways of managing airport operations, and are starting to deliver on their promise of real-time resource optimisation. Taken as a whole, passenger satisfaction itself is a key focus for airports and these tools can also be applied to its optimisation.
Those who have flown regularly will have experienced a rich variety of experiences in airports around the world, each experience subjective and dependent on a myriad of variables and touchpoints – connection times, queuing times, cleanliness, friendliness of staff, quality of retail and restauration… all play measurable parts.
Quantifying and understanding the drivers of passenger satisfaction is not easy, but it is clear that providing an efficient, safe, enriching – and now, hygienic and reassuring – experience is not only a competitive advantage for airports but also an enabler in restarting the mobility so crucial to our societies.
Applying machine learning techniques to these challenges has led to some promising results.
The advent of IoT technologies enabled the collection of a wealth of data; by implementing automatic triggers (a soap dispenser is almost empty, for example), by continuously monitoring key processes using tools such as computer vision (ideal for measuring queue lengths [3]) and through the real-time collection of customer feedback (eg satisfaction buttons [2]) or emotion detection).
This has enabled operational teams to move from plan-based paradigms, in which activities were organised according to a relatively fixed schedule, developed over time and based on experience and observation, to an alert-based paradigm where real-time conditions dictate activities. This had generally led to an improvement in efficiency and customer satisfaction [4] but creates challenges in terms of prioritisation and workload management.
Machine learning models allow us to go further and develop forecast-based paradigms, enabling the prediction of customer satisfaction as a function of a significant number of interrelated external variables. In real terms, this allows an airport to know in advance what their passenger needs will be, and to understand the key drivers of satisfaction (and dissatisfaction) at a given time. Airport teams can then proactively intervene, prioritising the most impactful tasks.
VINCI Airports in collaboration with the AI program by Leonard, has recently developed and is testing such models in two French airports, focusing initially on two aspects: cleanliness and queuing.
At Nantes-Atlantique Airport, a solution has been developed to leverage past, present (real-time customer feedback) and future data (flight plans, weather forecasts) to model customer behaviour and to predict passenger satisfaction over the coming week. Dissatisfaction peaks and their likely sources are identified in advance allowing operations to adapt their cleaning plans and improve overall customer satisfaction.
At Rennes Airport, machine learning has enabled not only the forecasting of passenger volumes but it also drives their processing. Again with Leonard, VINCI Airports has developed and trialled a “Queueless Airport” concept, which assigns every passenger an individual timeslot to pass through security. This reduces the time that passengers waste in queues, facilitates better social distancing, and allows airport operations to better regulate passenger flow and density between landside (before security) and airside (between security and the gate). Machine learning models ensure that passengers spend the minimum of time queuing while ensuring that they do not miss their flight.
Both of these projects demonstrate the potential of machine learning models to have a real impact on passenger satisfaction and create win-win scenarios where limited airport resources are focused at the right time and the right place on what is most important to our passengers.
Sources
- ASQ Best Practice Report, Airport Cleanliness, 2012
- https://www.skiply.eu/
- https://www.xovis.com/en/home/
- Kansai Airport, Press Release, 2019
