Agentenbasierte Fahrgastinformationssysteme für den ÖPNV

Passengers of public transport systems face significant challenges in accessing information due to the increasing complexity of transport services and information channels. The use of different modes of transport operated by various service providers leads to a highly fragmented information landscape in which timetables, disruption notices, and real-time information are distributed across numerous sources. Especially in cases of delays or cancellations, this makes it difficult to obtain reliable, complete, and up-to-date information. Furthermore, existing passenger information systems are often not tailored to individual user needs and provide insufficient support in terms of personalization and accessibility. Multi-agent systems offer the potential to address these shortcomings by aggregating information from heterogeneous sources in real time and presenting it in a consistent manner. Through autonomous information acquisition, such as conducting web searches, as well as the use of data analytics, relevant information can be efficiently consolidated and multimodal routes can be calculated. Interaction via natural language enables the flexible capture of individual preferences, such as preferred modes of transport or mobility-related limitations, and their immediate incorporation into route planning. Moreover, such preferences can be learned over repeated interactions and taken into account in the long term. Against this background, the question arises as to which technical and regulatory requirements an agent-based passenger information system must fulfill. In addition to aspects such as security, data protection, reliability, and cost, the selection of suitable language models, frameworks, and system architectures is of central importance. The aim of this work is to develop a general conceptual framework that enables the implementation of user-friendly and context-sensitive agentic passenger information systems for specific public transport scenarios.