The ELETA’s project, agreed in the 2016 TEN-T Days in Rotterdam to boost international rail freight, has concluded that more accurate ETA data of freight trains are possible by means of algorithms.

Once quantity and quality of data is ready for use, forecast methods using AI can achieve even higher accuracy rates than real-time information, according to the results obtained.

The project’s main goal seeked the participation of stakeholders through a web-based application where data is inserted, combined and calculated. In this sense, it has concluded succesfully, since it has been capable of gathering the majority of parties involved.

Even though results related to ETA predictions are promising, challenges regarding standardization, legal framework for data sharing and collaboration remain the principal obstacles to overcome.

“Where data is complete and available in good quantity and quality, machine learning is already improving ETA predictions”

The main inputs and contributions are:

• Artificial Intelligence or machine learning ETA forecast methods using historic train journeys and past delays, can substitute previous methods used in the industry and obtain higher accuracy rates.
• The project has brought together the majority of stakeholders of the rail freight industry in a digital platform where data is shared, trains are tracked and ETA are forecasted. However, it is necessary to solve different issues regarding collaboration in first place.
• Standardization arose as a big concern during the project. A unique train number is crucial for traceability and data sharing between stakeholders. This unique reference will be addressed in the near future.

The roadmap of the project is:

• The project will focus on completing the list of stakeholders and is expected a follow-up.
• Reliability among the rail freight sector is known to be doubtful. Therefore, critics came up with the need to be punctual and prevent the delays as well, not just aim to forecast them.
• As different stakeholders are involved, different events gain importance regarding visibility and traceability, since each party has different interests.
• Taking into consideration customer demands, for instance the knowledge of train arrivals, will continue to be part of the project, even after the completion of it.

[button url=»https://www.railfreight.com/technology/2019/11/06/artificial-intelligence-improves-estimated-time-of-arrival/» class=»» bg=»» hover_bg=»» size=»14px» color=»» radius=»0px» width=»0px» height=»0px» target=»_self»] More information [/button]