Improving efficiency of on-demand buses through machine learning

Public transport services in rural areas are often very limited. To offer a customer-centric alternative in these areas, on-demand bus systems are increasingly being established. Operators are confronted with the difficulty that the available resources are limited. Further, demand exceeds supply, is extremely random and therefore difficult to predict. Public transport providers are therefore faced with the challenge of using their budget efficiently in these systems.

As part of the project newMind, Prof. Fontaine and his research associate Simon Mader have investigated how efficiency can be increased through machine learning methods. Additionally, they were supported by Stefan Voigt of the Chair of Supply Chain Management & Operations. At this year's BAI.CON Conference, Pirmin Fontaine presented these results and discussed them with experts from research and practice. The findings show that 16% more requests could be served in the tested area through the use of AI. Further studies show that the introduction of the Deutschlandticket has a significant impact on demand, depending on the system structure.

In the subsequent panel discussion, the challenges and possible future developments in public transport were discussed.