International Transportation | Collection 2023 18 Autonomous shuttles in-Bavarian Bad Birnbach Research for an AI-supported future of public transport in-rural areas Autonomous mobility, Artificial intelligence, AI mobility, Barrier-free mobility, On-demand mobility, Reinforcement learning, Predictive demand Can artificial intelligence (AI) improve the use of autonomous minibuses? Shuttles that know when and where they are needed - that is the ambition in the KI4autoBUS research project. More efficient fleet control, including design and use for people with mobility impairments, is being developed and tested as part of the project. For the project, the autonomous buses in use in Bad Birnbach (Lower Bavaria) are being converted to be barrier-free and temporarily controlled in the background with innovative AI-software. Nicole Wagner-Hanl, Julian Wagner, Leandra Rüpplein, Thomas Huber T he idyllic municipality of Bad Birnbach in the Lower Bavarian district of Rottal-Inn has been a pioneer of autonomous mobility in Germany. Since 2017, DB Regio Bus has been operating autonomous shuttles from the French manufacturer EasyMile in regular service. Since May 2022, the regular service has been supplemented by an ondemand service (Figure 1). This service visits 20-virtual stops whenever there is a demand. As part of the research project KI4autoBUS, the Fraunhofer Institute for Material Flow and Logistics IML, together with its partners Q_PERIOR AG, DB Regio Bus, qdive GmbH, and FMS Future Mobility Solutions GmbH, is working on the further development of autonomous shuttles as part of a holistic mobility service in Bad Birnbach. The goal of the project is to develop an AI-driven software that optimally adapts the limited number of shuttles to the needs of the users, creating an integrated and efficient on-demand mobility service through Shuttle Service in Bad Birnbach. Photo: Nicole Wagner-Hanl/ Fraunhofer IML BEST PRACTICE Autonomous mobility International Transportation | Collection 2023 19 Autonomous mobility BEST PRACTICE targeted transportation planning and fleet optimization. To achieve this goal, Q_ PERIOR is working with its AI specialists from qdive to develop an algorithm that considers mobility needs across the entire public transport system through static and situational data and adjusts the offering flexibly. As a first result, it has been shown that Artificial Intelligence can learn most effectively using a reinforcement learning model that requires minimal data. Furthermore, the mobility app “Wohin-du-Willst” will be enhanced with simple and intuitive functions, such as profile information, to enable all users to access the mobility service. During the project, one shuttle will also be modified to accommodate mobilityimpaired individuals, meaning it will be equipped with additional components to offer barrier-free rides. The autonomous decision-making process of the AI enables optimal transportation planning tailored to the specific requirements of each individual user, especially for disabled persons or vulnerable user groups who face challenges in maintaining independent mobility. Fraunhofer IML is responsible for the scientific coordination of KI4autoBUS. They conduct the requirements analysis, usability and acceptance evaluations, and are involved in the transferability of project results and scientific exploitation. Conclusion on the use of AI Added value for mobility providers and travelers? - An innovative component for optimizing public transport? For the optimization of on-demand shuttles in Bad Birnbach, modern technology using AI has been chosen, which can help save resources and enhance new forms of mobility. The self-learning algorithm is intended to ensure that the offering can flexibly and quickly respond to various user needs. The optimization through AI offers advantages for users by avoiding long waiting times through predicting ride requests or determining the most convenient waiting stop. At the same time, mobility providers benefit by minimizing empty trips, optimizing fleet size, and reducing operating costs through improved shuttle utilization. From the perspective of mobility providers, the entire public transport system can benefit from AI-driven fleet and traffic planning. When flexible services such as ondemand mobility complement existing public transport offerings into an integrated overall service, and this service becomes increasingly automated in the coming years, self-learning systems will be necessary to be efficient and user-oriented. Complex and automated mobility services, particularly in rural areas, require the support of AI technology. Applied research and user-centered development have shown that mobility will increasingly adapt to the individual needs of travelers in the coming years, for example, through new offerings based on autonomous vehicles in public transport. The foundation of these mobility services relies on real-time data about traffic, traveler demand, and the availability of interconnected services. The secure processing and analysis of this data in distributed systems will be crucial success factors for the mobility of tomorrow. ■ Leandra Rüpplein, M.Sc. Consultant for digital public transport solutions, Q_PERIOR AG, Munich (DE) leandra.ruepplein@q-perior.com Thomas Huber, Dr. Leiter Innovative Verkehrskonzepte, R.RS-BY-VV, DB Regio Bus, Ingolstadt (DE) thomas.ta.huber@deutschebahn.com Julian Wagner, Dr. rer. nat. Senior Data Scientist, qdive GmbH, Munich (DE) julian.wagner@qdive.io Nicole Wagner-Hanl, M.A. Consortium management KI4auto- BUS, Project Manager Mobility and Digitalization, Fraunhofer Institute for Material Flow and Logistics IML, nicole.wagner-hanl@iml.fraunhofer.de Figure 1: Route network autonomous driving Source: DB Regio Bus The project is funded by the Bavarian Ministry of Economic Affairs, Regional Development and Energy (StMWi) as part of the Digitization funding program, in the Information and Communication Technology funding area, and is expected to run until the end of 2023. LITERATURE KI4autoBUS: Optimization of Barrier-Free Mobility through Autonomous Shuttles - Development of an AI-Based Solution for Planning and Control of Public Transport Services. www.iml.fraunhofer.de/ de/ abteilungen/ b3/ Projektzentrum_Verkehrslogistik_Prien/ Referenzen/ kl4autobus. html, 03 July.2023