Radix and Brussels Airport company revolutionise luggage retrieval

Case study

The challenge

Brussels Airport Company (BAC) has been moving ahead very quickly with its digital transformation. As more and more people take to the skies, airports across the globe are faced with the challenge of accommodating an increasing number of travellers. Airports need to find ingenious ways to efficiently process tens of thousands of pieces of luggage every day.

Brussels Airport is the biggest international airport in Belgium. In 2019, the airport welcomed more than 26 million passengers and operated 200.000 passenger flights. More than 41 thousand suitcases and pieces of luggage are handled every day at Brussels Airport.

The briefing

Brussels Airport seeks to make the customer journey as smooth as possible for travellers arriving from across the globe. Both arriving and departing travellers are looking for a pleasant, quick, and overall hassle-free experience.

The collaboration between Radix and BAC started with a simple, but crucial observation: luggage retrieval can often be a source of stress for passengers during their trips. One of the challenges airports face is to accurately inform arriving passengers about the status of their luggage so they can make the best use of their time.

BAC & Radix started working together on a tool that could inform passengers on the time they can pick up their luggage, to help make the experience stress-free. The solution needed to be:

The work

Radix used AI and machine learning, together with BAC’s internal team, to make predictions on luggage retrieval time for passengers. The project officially started in June 2018.

The exploration phase

The project started with an exploration phase. Radix scanned useful, available and actionable data to build its machine learning model: flights, airlines, airport of departure, number of passengers, weather conditions, luggage carriers, bag specifics, … Using this data, Radix developed a unique algorithm capable of predicting when a piece of luggage will become available for pickup.

The prototype

The second phase of the project was dedicated to Radix’ first prototype. Radix built a prototype machine learning model to test its predictions, just 4 months after the beginning of the collaboration. The process was automated: when a new flight arrives at Brussels Airport, BAC’s database sends an automatic trigger to the machine learning model. The prototype then makes an exact time prediction of the luggage’s time of arrival on the reclaim belt, taking into account where the flight is coming from, the time of the day and several other conditions. This prediction is then communicated to the passenger along with the walking time from the plane to the reclaim area and the specific belt the luggage will be on. The information is automatically sent to the passengers: they know when to retrieve their luggage as soon as they leave their flight.

To make the necessary calculations, the machine learning model uses a serverless architecture. The predictions are made using computational resources in the cloud, when the model needs them. There are multiple advantages to this specific architecture: first, there’s no need to provide and maintain computing services. The architecture is also cost-effective, as it only charges for the computational resources actually used by the machine learning model. When more predictions need to be made, this architecture makes the operation easily scalable.

Keeping the end user in mind

Radix took the passengers into account during the whole process to develop a practical and user-focused tool. The prototype was tested by customers in order to get feedback at the end-user level. It was an essential part of the project, allowing Radix and BAC to bring to light some elements that might’ve gone unnoticed at first glance. One could assume that passengers would want to be in the retrieval area a bit before the first bag starts arriving.

However, feedback from passengers showed that most passengers would rather simply arrive and pick up their bags. Radix integrated this consideration into the model to better meet passengers’ expectations. The model provided travellers with an accurate and custom estimate of their luggage’s time of arrival, contributing to a stress-free arrival and giving them the opportunity to use their time more efficiently.

Our agile approach

Radix’ approach entailed implementing the solution fast and learning from feedback and results in a very agile way to make the best possible model. This allowed Radix and BAC’s internal team to very quickly start testing the model, without needing to develop beforehand a completely standardised solution and design – which is a long process. This approach enabled direct integration of end users’ feedback into the solution, keeping those end users at the centre of the process.

A tangible impact

Radix’ solution had a direct and considerable impact. 95% of its predictions are accurate, with 90% of passengers rating the tool “excellent” or “very good ”. 35% of passengers also reported a change in their behaviour (going to the toilet, buying a drink, …) due to the increased freedom of mind allowed by the device, permitting them to make better use of their time at Brussels Airport.

The future

The collaboration of BAC and Radix created a real impact on passengers’ experience at Brussels Airport. This was made possible by finding the right AI opportunities and leveraging them to improve customers' lives, while actively integrating feedback from the end users. This case was the first use of cloud and serverless architecture at BAC, as well as the first-ever use of machine learning.

The model keeps getting better by learning from new data and is integrated in BAC’s bTag. The positive contribution of AI and machine learning to BAC’s workflow has also opened the door for additional innovations and use cases from Radix at the airport.