Development of an AI-based decision support system for plannable or short-term disruptions in public transport operations

The background

The Freiburger Verkehrsbetriebe (VAG) operates streetcars and buses in
buses in Freiburg and the surrounding area. The operation of these means of transport is highly dependent on external conditions in the urban traffic area. External influences such as traffic accidents, roadworks, short-term road closures or major events can affect regular operations and cause restrictions in the route network. Short-term, unforeseen situations such as traffic accidents, spontaneous road closures or first-time events require quick reactions in order to keep restrictions on the route network to a minimum. This contrasts with plannable events, in particular roadworks. These are usually announced in advance and require early coordination and suitable operational measures. In both cases, well-founded decisions must be made in the operations control centre, for example to reroute routes, set up replacement services or adjust frequencies in order to minimize the impact on operations and passengers.

The aim of the cooperation is to overcome these operational challenges more efficiently using modern methods of artificial intelligence and machine learning. The analysis and evaluation of historical data volumes form the basis for learning processes. These are used to ultimately develop an AI-based assistance system that supports decision-making processes in the control center based on data, provides recommendations for action and supports resource planning. The master's thesis is a collaboration between KIT, FZI Research Center for Information Technology and VAG.

Objectives

• Analysis of historical data on disruptions, construction sites and adaptation measures carried out in VAG bus and streetcar traffic

• Conception and simulation-based investigation of adaptation strategies for defined plannable or short-term disruption scenarios

• Evaluation of the identified adaptation strategies

• Development of an AI-based decision support system

• Application and evaluation of the model using real scenarios

• Documentation of the work and presentation of the results on site

Prerequisites

• Interest in mobility and public transport processes

• Basic knowledge of data analysis, AI methods and machine learning processes

• Experience in modeling and simulation

• Structured way of working

Benefits

Gain relevant practical experience, opportunity to travel regularly to Freiburg, state-of-the-art AI methods and machine learning processes in application