Intelligent Systems for Urban Planning

Data-driven Solution for the Smart City Hamburg

D²S²C Hamburg - Prototypes & Results

In the project "Data-driven Solutions for the Smart City Hamburg" (D²S²C Hamburg) , students of the MIN faculty have the opportunity to work on different use cases in groups in a practice-oriented environment to develop intelligent prototypes with the help of innovative technologies such as machine learning.
In this process, the students are supported by the lecturers and cooperation partners from business and public administration, who provide data and information for this purpose and are part of the agile development process. The seminar is initiated and led by Marten Borchers from the AG WISTS at the University of Hamburg and accompanied by Simon Jordan. In the summer semester of 2022, the seminar successfully collaborated with HOCHBAHN, HSV/Future Dock, and the State Office for Geoinformation and Measurement, and four use cases related to Smart City were elaborated.

The four prototypes for the 2022 summer semester are presented and briefly described below.

Analysis of weather-related Causes of Incidents to support Reporting

Project Picture for Identification of Accident Hotspots — University of Hamburg

Lecturers: Marten Borchers & Chikaodi Uba
HOCHBAHN: Daniel Schulz & Jan Krause
Students Summer Semester 2023: Lieven Petersen (ORCID) & Nils Birker (+2)

Due to climate change, the demands on transport infrastructure and its resilience are changing. The focus is on weather-related factors, including events such as heavy rain, flooding, snow, black ice, and storms (e.g., falling trees), but also roads that are temporarily overloaded due to high traffic volumes at peak times or due to major events, construction sites, and accidents. These factors should be considered when planning the route network as well as urban development (soil compaction and alternative surfaces).
To identify possible influences on the operation of HOCHBAHN, a prototype application for the analysis of disruptions was developed together with HOCHBAHN. For this purpose, exemplary selected operational reports were analyzed and an AI-based system for the identification of relevant events was developed. These were supplemented with data from a weather database to visualize possible weather-related disruptions and trends and to identify correlations using filter functions (e.g., by time, period, day of the week, weather event), which can then be checked manually. Insights gained can thus be used to develop solutions to increase the resilience of the transport network as well as to raise awareness of possible consequences of climate change.

Intelligent Stakeholder Analysis and Management

As the central mobility service provider of the city of Hamburg, HOCHBAHN is involved in many changes and developments, which include the expansion of required infrastructure. During these projects, the HOCHBAHN interacts with numerous stakeholders or committees and public stakeholders (e.g. politicians, committees, public associations) directly or in the context of public participation.
To support the participation and management of stakeholders, it was analyzed how interested and relevant stakeholders could be identified and included. For this purpose, a prototype was developed to investigate how stakeholders and websites can be identified and analyzed automatically with the help of a web crawler. In the process, various challenges emerged. Among other things, these concern the selection of websites, since a targeted and possibly also site-specific search leads to dozens or hundreds of results including irrelevant content. In addition, determining potentially interested stakeholders is difficult and can only be identified in an automated way to a limited extent using publicly available information, also because websites are designed and structured differently. An AI-based solution should therefore be able to recognize the basic structure of web pages and interests in the context of mobility expansion. The short details and different expressions make this difficult, as semantic comparisons are usually more reliable due to the density of information in longer texts. The prototype illustrates important insights that could be gained about the use of web crawlers and AI-based evaluation.

Lecturers: Marten Borchers & Chikaodi Uba
HOCHBAHN: David Claus & Jan Krause
Students Summer Semester 2023: 6

Analysis of Sustainability Reports of Public Transport Companies to Identify Activities and Measures

Lecturers: Marten Borchers & Chikaodi Uba
HOCHBAHN: Daniel Schulz & Jan Krause
Students Summer Semester 2023: Archit Sharma (+4)

The transformation to sustainable and environmentally friendly transport is already in full swing and is shaped at the European level by the European Green Deal and in Hamburg by the Hamburg Mobility Strategy (ITS). To achieve this, a wide variety of technologies (e.g. e-buses, autonomous driving, hydrogen) and concepts (interaction of buses, trains, bicycles, e-scooters, etc.) are being used. The effects and benefits are difficult to predict and quantify, as they also depend on local geographic, social, and economic conditions. Nevertheless, this does not only concern Hamburg or HOCHBAHN but all European mobility providers.
To identify activities, measures, and results that have been and are being taken in the context of the transformation as well as the comparability of these, existing sustainability reports were analyzed and evaluated in cooperation with HOCHBAHN to determine information requirements and to develop an intelligent tool for the automated recognition and extraction of these. This tool is based on the Python programming language and integrates the generative AI system of Open AI as well as specially developed machine learning models to recognize and extract the desired data accurately and reliably and to present it in a graphical user interface.

Identification and Visualization of Accident Hotspots

With more than 1,100 buses, the HOCHBAHN is Hamburg's central mobility provider and a crucial part of the public transportation system. With such an extensive fleet, accidents occur in interaction with other traffic players and are favored by road works, constructions, and varying weather conditions.
In cooperation with HOCHBAHN, the project is investigating whether and how recurring patterns in accidents can be identified to take measures to further improve the safety of bus traffic. For this purpose, discussions were held with representatives of HOCHBAHN on how information on past accidents, collisions, and property damage could be processed and made usable for a systematic analysis. The analysis and exchange showed that a visual representation is most intuitive for the analysis of accidents and the identification of patterns by experts. The knowledge gained from the research project was used to develop an interactive web prototype using Flask and JavaScript, which enables the exemplary display of data. OpenStreetMap is used to display the city of Hamburg and surrounding regions and was supplemented with locations of exemplary accidents. These can also be displayed or hidden according to various criteria such as weather (e.g., light conditions, rain), time, day, and line numbers.
The prototype demonstrates the advantages of a visual representation of incidents. Nevertheless, it becomes apparent that the system can only be used if incidents are reliably and consistently assigned to locations. These and other technical as well as user-oriented aspects are and will be further investigated to support HOCHBAHN experts in the future.

Lecturers: Marten Borchers & Simon Jordan
HOCHBAHN: Reinhard Hübener & Jan Krause
Students Winter Semester 2022/23: Berk Güngör & further students (+4)
Students Summer Semester 2022: Larissa Rutzen , Amar Talib & further students (+1)

Prediction of the Energy Consumption of E-Buses

Project Picture for Energy Prediction of electronic Buses — University of Hamburg

Lecturers: Marten Borchers & Simon Jordan
HOCHBAHN: Johanna Ahrens & Thilo Kemper
Students Winter Semester 2022/23: Jan Willruth & further students (+3)
Students Summer Semester 2022: Niklas Beneke & Hendrik Koser

The number of electrically powered buses at HOCHBAHN is successively increasing. The aim of forecasting the energy requirements of electric buses is to support operations by providing adapted and reliable energy demand information. This is intended to use battery capacities as efficiently as possible, reduce charging cycles and optimize vehicle deployment. For this purpose, several machine learning models were trained using circulation data. A round trip describes the travel time of a bus and can include several lines. The attribute to be determined is the energy demand for a specific circulation considering the different vehicle types/battery types. The amount of energy required depends on many different attributes, such as the distance to be traveled, the bus model, the line, the time of day, the outside temperature, the traffic density, and the nature of the route. The quantification of the traffic density was determined depending on the start and end time and the traffic data of the Urban Data Platform Hamburg. In addition, weather-related information was queried via weather API and numerous machine learning models were trained to determine and compare both the influence of the different features on the prediction and the precision and reliability of these.

Intelligent Station Navigations

In the use case of intelligent navigation in the Volksparkstadion of the HSV ( HSV/Future Dock), it was investigated how viewers can be navigated through the stadium by using the shortest route to their seats. As a basis for this, a 3D model based on architectural plans of the stadium was created using computer vision. For this, the plans of the different levels were transformed into graphs and the contrast was increased to sharpen the structures to highlight seats, entrances, exits, and stairs. Then, doors and texts were detected using machine learning models to lay a grid over the walkable area that allows navigation with nodes and edges. The graph is constructed to be easily adaptable, to allow support for different events for which the navigation has to change as part of the stadium is not accessible for example in the case of concerts. For the application, the 3D model was connected to an interface that was developed with python and HTML. The intelligent navigation system can be used by entering the seat location and Dijkstra's algorithm which calculates the shortest path which is visualized in the interface. The starting point is one of the stadium entrances, which needs to be replaced by the location of the viewer in real-world facilitation. In perspective, an extension of the application can also consider stopovers at food or merchandise counters and take care of an even distribution of visitor flows with the help of the GPS function of mobile phones to reduce waiting times and crowded corridors.

Lecturers: Marten Borchers & Simon Jordan
HSV/Future Dock: Julia Hildenbrand
Students Summer Semester 2022: Jan-Gerrit Habekost, Robert Johanson & Paul Starke

Project Picture for Smart Stadion Navigation — University of Hamburg

Intelligent Analysis of Citizen Contributions from Urban Participation

Project Picture for Citizen Participation - DIPAS — University of Hamburg

Lecturers: Marten Borchers & Simon Jordan
Agency for Geoinformation and Surveying: Bianca Lüders
Students Summer Semester 2022: Dmitri Milovanov, Tim Radke, Hagen Peukert & further students (+1)

In the use case from the State Office for Geoinformation and Measurement, the automated evaluation of citizen contributions originating from participation projects of the digital participation system DIPAS was considered with the aim of supporting the evaluation of participation data and accelerating it to especially support large participation projects. The data e.g. contributions from citizens are first preprocessed and classified into predefined classes. Based on this, different models were trained to classify new citizens' contributions as reliably as possible to facilitate the structuring of the content. For this purpose, Naïve Bayes (Figure Left - left) and a Bert classifier (Figure Left - right) were implemented and evaluated. In addition, approaches for summarizing groups of contributions to extract the most important information from the texts in the form of key - phrases were implemented. Special attention was paid to the transparency of the algorithms and machine learning models to make the decision support as comprehensible as possible. The insights gained from processing and summarizing the citizens' contributions provide information about the requirements and wishes of citizens, which can support urban planners and contribute to the design of the urban living space.

Template by OS Templates

Information on the use of cookies