Arrival baggage prediction service
Incheon Airport
Can passengers receive more accurate information about their baggage before it reaches the carousel?
This proof of concept explored how predictive analytics can improve the arrival experience by providing passengers with real-time information about baggage delivery and claim progress.
Challenge
After completing immigration procedures, passengers arriving at the baggage reclaim area receive very limited information about the status of their baggage.
This lack of visibility can create uncertainty and frustration, particularly during waiting periods. The challenge was to determine whether historical and operational baggage data could be used to provide passengers with reliable arrival predictions and real-time progress updates
Approach
A baggage prediction service was developed using historical flight and baggage handling data. Machine learning models were trained to predict:
• First bag arrival time
• Last bag arrival time
• Baggage claim progress
The predictions were displayed on baggage reclaim screens, allowing passengers to receive more detailed information throughout the baggage delivery process.
"Passengers do not just want their baggage quickly—they want to know when it will arrive. Better information can be just as valuable as faster delivery." – Juhyang Choi
Assumptions
We believe that:
• Predictive analytics can provide passengers with more accurate baggage information.
• Better information can improve the passenger experience during baggage reclaim.
• Historical operational data can be used to predict baggage delivery milestones with high accuracy.
• Prediction services can be integrated into existing airport information systems.
• Real-time baggage visibility can reduce uncertainty and improve passenger confidence.
Key learnings
Product – accuracy drives trust
The proof of concept demonstrated the potential of machine learning to predict baggage delivery milestones using operational and historical data.
Further model optimisation is required to achieve the target prediction accuracy of 90% or higher and ensure reliable performance across different operational scenarios.
People - Passenger value comes first
The technology showed promise, but its success ultimately depends on whether the information provided meets passenger needs.
Future testing will focus on passenger feedback, while exploring opportunities to combine prediction services with baggage tracking information to create a more transparent baggage reclaim experience.
Process – prediction only valuable when operationally integrated
Defining prediction cycles, system triggers and data exchange proceses is very important.
Successful deployment depends not only on model performance, but also on seamless integration with the airport systems that power passenger information displays and applications.
.jpg)
Looking ahead
The proof of concept confirmed that AGVs can play a meaningful role in the future of baggage handling. While further development is needed to improve robustness and integration, the direction is clear: autonomous transport has the potential to support safer, smarter and more resilient baggage operations.
The next step is not proving the concept works—but making it robust, integrated and scalable enough for everyday operations.
Juhyang Choi
Head of Baggage Brussels Airport
Let's connect
Interested in the lessons learned from this proof of concept?
Get in touch to discuss the findings, challenges and opportunities for AGVs in baggage operations.