AI can already provide faster and more accurate forecasts compared to a traffic manager, when it comes to shorter disturbances in train traffic.
This is demonstrated by an ongoing research project in which the Swedish Transport Administration, Linköping University and Atea are collaborating on.
But it is in the interaction between man and machine that the greatest values will arise.
The aim of the research project is to show how the Swedish Transport Administration can deliver more reliable forecasts and more accurate answers about when traffic is expected to resume, following a disruption in train traffic.
The goal is to train the algorithm so that we can more quickly send out more accurate traffic forecasts for all kinds of traffic disruptions.
The AIRT project, AI-based Real-time forecasting of Traffic Information, has reached the halfway point. In the first year, the research team was designed and data sources were evaluated. Algorithms were written, tested and trained, and data was prepared and entered into the database.
To make the tasks "data ready", people are needed who can analyze and understand the Swedish Transport Administration's disruptions, but also levels (of traffic disruptions) and reason codes. At the same time, the sharpest AI experts are required. Therefore, the Swedish Transport Administration chose to hire Peter Lenaers, AI expert at Atea, who collaborates with researchers from Linköping University.
The goal of the research project is to show how the Swedish Transport Administration can deliver safer forecasts and more accurate answers about when traffic is expected to start again, after a disruption in train traffic.
“_We have seen a number of exciting results so far, but we have almost a year left, so much can happen. In December 2022, we will present what we have arrived at and after that a decision will be made about a possible continuation or implementation_” says Daniel Jakobsson, Strategist at Trafikverket
