Prof. Dr. Martin Raubal Gaze-Based Interaction for Touristic Urban Spaces for Multiple Users

Transcription

1 Contents Gaze-Based Interaction for Touristic Urban Spaces for Multiple Users... 2 Identification of potential ride-sharing paths from GPS taxi trajectory data... 3 Beyond Pokemon Go: successful (re)localization of location based games for crowdsourcing applications... 4 A Holographic Campus Pedestrian Navigation System... 5 Open Communication by Augmented Reality Experiences in Urban Planning... 6 What is a valley?... 7 Construction of a user-level probabilistic forecast of electric vehicle usage... 8 Implicit Calibration for Gaze-Based Interaction with Maps Using eye tracking for improving tactile feedback in an aviation task Gaze-Based Assistance for the Color-Blind The Future of Logistics: How will Autonomous e-trucks Running on Renewable Energies Shape Future Energy Demand and Distribution... 13

2 Tiffany CK Kwok Gaze-Based Interaction for Touristic Urban Spaces for Multiple Users Exploring a city panorama view from a vantage point is a popular tourism activity during a city visit. Audio tourist guides are commonly adopted as a tool for introducing the city view. They can provide information to a tourist without disturbing her visual field. In recent research, it has been considered to adapt the content of touristic audio guides to a user s gaze, thus taking the tourist s interest and pace of listening into account (see LAMETTA, However, the system only enhances the interaction between the human and the environment without considering the type of human-human interaction naturally occurring when tourists visit a destination in a group. In this thesis, the student will develop and evaluate a multi-user gaze-based audio guide for tourists exploring a city panorama. Challenges to be addressed include: designing a gaze-assisted interaction method between two or more individuals and designing a study to evaluate usability and user experience. Particularities (e.g. comments on group work etc.): Requirements: knowledge in an object-oriented programming language - Java (e.g., acquired through the Master courses Mobile GIS and LBS or GIS and Geoinformatics Lab ). no 2

3 Dr. Pengxiang Zhao Dominik Bucher Henry Martin Identification of potential ride-sharing paths from GPS taxi trajectory data As an effective way of alleviating urban traffic congestion, reducing traffic pollutant emissions and energy consumption, ride-sharing has attracted considerable attention from governments, enterprises, and scholars all over the world in recent years. Identifying the potential paths suitable for sharing is one of the key components of ride-sharing. rmally, occasional human trips are inconsistent or even quite different in time and space. In addition, influenced by uncertainties such as traffic congestion, it is difficult to guarantee the arrival time of passengers. Hence, it is necessary to identify the potential ridesharing paths. Extensive GPS taxi trajectory data availability and associated processing technologies offer an efficient way to identify ride-sharing paths. The taxi trajectory data records the traces of taxis during driving in detail, including geolocations, time, speed, direction and passenger status. Passengers trips are represented as trajectories restricted by the road network. The potential ride-sharing paths are similar in time and space. Through measuring the spatiotemporal similarities of the corresponding trajectories, the trajectory clustering method is used to detect the path clusters, thereby achieving effective identification of the ride-sharing paths. Particularities: The goal of this thesis is to develop a framework for the identification of the potential ride-sharing paths from GPS taxi trajectory data. The student will first review the related literature on ridesharing and trajectory clustering. A more specific introduction into the topic will be given by the advisor upon request. Interested in data analysis and mining. Knowledge of Python is beneficial. 3

4 Dr. Simon Scheider (Universiteit Utrecht) Beyond Pokemon Go: successful (re)localization of location based games for crowdsourcing applications An increasing amount of geoinformation comes from volunteers who share this information on collaborative platforms, such as Open Street Map (Volunteered Geographic Information, VGI). In order to motivate contributors and to increase the quality, spatial coverage and up-to-dateness of the data, it has been proposed to use location-based games (LBGs). These games are played by one or several players who move in geographic space and exchange (virtual) resources at physical places. While LBGs have been around for a while, only few of them have succeeded in attracting a larger number of players. Pokemon Go is one of the most recent and most successful examples. Particularities: One reason is the dependency of game concepts on particular locations. In order to reach players from different places, LBG concepts need to be relocalized in a way which preserves the particular attractiveness of a game. What are criteria for high-quality relocalization? While the problem has been recognized in the literature, a systematic derivation of corresponding criteria is missing. The goal of this thesis is to investigate the conditions for a successful relocalization of a LBG regarding three conceptual levels: the ludic level (game rules and roles), the narrative level (story of the game), and the environmental level (real-world places). As an example, the student will develop his or her own game (running as an app) and study its relocalization. A location-based game based on OSM street network data, student project Universität Bamberg, 2010, 4

5 Fabian Göbel David Rudi A Holographic Campus Pedestrian Navigation System In recent years, head-mounted interfaces for mixed and augmented reality have become increasingly robust and affordable. These types of interfaces enhance the real world with visual overlays, providing their users with an immersive user experience and intuitive interaction. A particularly interesting device for the interaction with location-based information is the Microsoft HoloLens ( a holographic interface that determines and uses a 3D model of its environment to visualize 3D objects in near space. The goal of this thesis is to develop a pedestrian navigation system for (part of) the Hönggerberg campus on Microsoft HoloLens. Challenges to be addressed include: positioning of HoloLens in an environment with indoor, outdoor, and transitional spaces; making a pedestrian routing service accessible from HoloLens; designing a holographic user interface that communicates pedestrian routes in an efficient and effective way. A usability evaluation of the system will be performed. Wikimedia, Creative Commons, Author: Code Teacher Particularities: Requirements: knowledge in an object-oriented programming language (e.g., acquired through the Master courses Mobile GIS and LBS or GIS and Geoinformatics Lab ). 5

6 Prof. Dr. Martin Raubal Christian Sailer Matthias Schenker Esri Schweiz AG Christian Hürzeler Amt für Städtebau der Stadt Zürich Open Communication by Augmented Reality Experiences in Urban Planning Over the past 20 years, the population of the city of Zurich has grown by more than 50,000 people and will increase by another 100,000 in the next 20 years. In order to create a living and working space for the increasing number of people in Zurich, a sustained high level of construction activity is expected to continue in the future. As a pioneer in the field of Smart City, the city of Zurich has set itself the goal of involving the population as much as possible in the urban planning process. A new option for communication is Augmented Reality (AR). As part of this Master's thesis, a prototype AR app will be developed that enables the visualization of planned construction projects on-site on citizens own mobile devices. Research questions How suitable is an AR application for visualizing new building projects in an understandable and realistic way, compared to conventional methods such as wooden construction spans or physical models? How difficult is it for users of an AR app to interpret the visualization correctly? Prototype / technological framework The prototype will be implemented with the AR / VR ArcGIS Runtime and will use 3D services of the city of Zurich. In contrast to conventional visualization methods (construction span, physical models, photomontages), the app allows on-site visualization of new construction projects and analysis (visibility, shadow, etc.). Positioning with more accurate procedures than the smart phone s GPS sensor (such as markerbased or image recognition) Further technological framework (e.g., use of ios with ARKit 2.0): Particularities: Important skills are: General programming skills (ArcGIS Runtime SDK) Designing a user study no 6

7 Joram Schito What is a valley? While hiking or walking around we easily recognize big geomorphological structures as e.g., valleys, hills, or mountains. Maps can further support us in recognizing them. However, in Geocomputation this is more difficult. Even though some approaches exist, it will be interesting to investigate how far novel approaches such as Machine Learning can be helpful to determine valleys more accurately. In this thesis, you will focus on the Alps and develop different models to determine primary valleys that are feasible for building new linear infrastructure such as roads, railways, pipelines, or power transmission lines. As a goal, you will work with different data models and compare the resulting approaches quantitatively with each other. In order to validate your model, you will have the opportunity to conduct a user study with representatives of various electricity companies. Particularities (e.g. comments on group work etc.): A more specific introduction into the topic will be given by the advisor upon request. Knowledge in R or Python is beneficial. 7

8 Henry Martin Dominik Bucher Manuel Lösch Construction of a user-level probabilistic forecast of electric vehicle usage Electric vehicle (EV) users often charge their EVs at home. By doing so, EVs can offer flexibility in the charging process to power markets, or provide grid services to the distribution system operators (DSOs). The possibility of using EV batteries on a large scale would not only have a strong positive effect for the overall power grid stability, but may also lead to financial rewards for the EV owners. Independent of the financial incentives, for a user, it is most important that her car is always available to serve her mobility needs (e.g. it must not happen that the user wants to go shopping but the EV is not sufficiently charged). A possibility to utilize the flexibility in charging EV batteries without threatening its availability to the user is that the user plans her trips ahead. When knowing about future trips of the user and therefore the required energy for the EV, the charging can be scheduled according to power market prices or grid needs. However, in practice, the user will often forget to declare her trips beforehand, which might result in situations where the EV is needed but not available. A solution to this problem is to optimize the EV s charging schedule using a probabilistic forecast of the user s mobility. Particularities (e.g. comments on group work etc.): In this thesis, the primary goal is to construct a probabilistic forecast of the EV usage. To this end, recent literature on mobility forecasts has to be studied, common methods for probabilistic forecasts (e.g. quantile regression, copula models or neural network based forecasts) have to be researched and assessed for their suitability to solve the problem. Based on this, a suitable probabilistic forecasting method has to be chosen and applied to mobility data (trajectory data of EV users will be provided). Finally, the probabilistic forecast has to be evaluated using suitable metrics (e.g. CRPS = continuous ranked probability score). Interest in: Data analysis, Probability, Programming (e.g. Python) no 8

9 9

10 Fabian Göbel David Rudi Implicit Calibration for Gaze-Based Interaction with Maps Gaze is a valuable resource for both analyzing a user s attention and for quickly and naturally interacting with a system. In order to utilize eye tracking, either in analysis or interactive manner, a calibration is required to correctly map the detected gaze with the actual focused point. We can differentiate theeyetribe.com between two calibration methods: Explicit and implicit procedures. During explicit calibration, the user will be provided with targets that s/he has to focus on one after another. This way, the eye tracker can learn the specific characteristics of the user s eye. While this method can be very precise, it may feel cumbersome in an interactive system, as a user has to perform the calibration prior to actually controlling the system with gaze. Another approach is to implicitly calibrate the system during interaction. Due to the nature of this approach the calibration can be integrated seamlessly with an interaction system. However, at start it delivers a low accuracy that steadily improves during interaction. Particularities (e.g. comments on group work etc.): The goal of this work is to compare different explicit and implicit eye calibration methods in regard to robustness and seamlessness. Based on this, an implicit calibration procedure for gazebased interaction on maps should be developed. The chosen design should be evaluated with a small user study. A more specific introduction into the topic will be given by one of the advisors upon request. Knowledge in Java and C# is beneficial. 10

11 David Rudi Using eye tracking for improving tactile feedback in an aviation task Vibro-tactile feedback has been used in different scenarios, such as to provide navigation directions or to simulate haptics for remotely operating surgeons. While vibro-tactile feedback has also been used to guide car drivers or pilots attention in a given direction or to provide collision information, it has not yet been utilized in combination with eye tracking to support the vibro-tactile feedback loop. In this thesis, the student will develop a wearable device, which is used in combination with eye tracking and a flight simulator to provide pilots with vibrational cues enhancing the flight experience. Particularities (e.g. comments on group work etc.): Requirement: Knowledge of Android programming, as taught in the Master course Mobile GIS and LBS. 11

12 David Rudi Fabian Göbel Gaze-Based Assistance for the Color-Blind Color-blindness or color vision deficiency is the decreased ability to see color or differences in color. Color vision deficiency affects approximately 1 in 12 men (8%) and 1 in 200 women in the world. There have been different approaches on how to assist those affected. The solutions primarily differ with regard to the feedback modality. For example, a user s visual perception of an item was manipulated as to highlight a particular color using head-mounted displays. Moreover, for one approach, a user received color-coded vibrational cues that were given to her using a wristband, while for another solution the user was given simple audio feedback. That is, the recognition of the object of concern is based on image recognition approaches that extract color information from a camera scene and can therefore become imprecise. Figures from: (LEFT) Enrico Tanuwidjaja et al., Chroma: a wearable augmented-reality solution for color blindness, Proceedings of the 2014 ACM International Joint Conference on Pervasive and Ubiquitous Computing; (RIGHT) Marta G. Carcedo et al., HaptiColor: Interpolating Color Information as Haptic Feedback to Assist the Colorblind, Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems] Particularities (e.g. comments on group work etc.): In this thesis, the student will use eye tracking technology in combination with basic image recognition algorithms to allow for a more precise identification of a user s object of interest. Furthermore, the student will conduct a user study to evaluate this novel input modality in combination with visual, tactile and audio cues as feedback techniques. Requirement: Knowledge of Android programming, as taught in the Master course Mobile GIS and LBS. 12

13 Dominik Bucher René Buffat Henry Martin The Future of Logistics: How will Autonomous e-trucks Running on Renewable Energies Shape Future Energy Demand and Distribution Several large car manufacturers recently announced their plans of mass-producing electric trucks in the near future (Tesla, Mercedes, Volvo, MAN, etc.). While many of them only exist as prototypes boasting with enormous range and acceleration values, we can still expect the first of them to hit the roads in the coming years. As transport and mobility are responsible for large shares of CO 2 emissions, it is of particular importance to increase the ecological sustainability in this field. While the CO 2 emissions of individual mobility (e-cars) are often researched, it is much less clear for the transportation of goods. But similar questions must be asked for logistics and transport: How much (renewable) energy will be required to power the logistics sector? Where will we have to provide this energy? Are there enough renewable energy sources close by or do we need to transport energy over large distances? Particularities (e.g. comments on group work etc.): In this thesis, the student will build a GIS-based model of the distribution of goods using trucks within Switzerland, and the related energy requirements. Using data on the potential production of various sources of renewable energy (solar, wind, water), the model will be used to quantify the energy demand from the transport industry potentially coverable by renewable sources in various scenarios (e.g., an increase in solar or wind power production). Finally, the applicability of the model to other geographic areas will be discussed and relevant differences highlighted. Interest in: Data analysis, modeling Experience with Python beneficial 13