In the Project Development Arena, students take the initiative in solving social problems.

AY2023-2024 Project List
  No.    Project Instructor
23-01 Smart Museum YOSHIOKA Rentaro
23-02 High-speed autonomous driving on low-friction road surfaces OKUYAMA Yuichi
23-03 Smart Learning WATANOBE Yutaka
23-04 Smart Information Visualization TAKAHASHI Shigeo
23-05 Analytical Tools for Network Science HASHIMOTO Yasuhiro
23-06 Analysis of Infodemic on Social Media HASHIMOTO Yasuhiro
23-07 Extensive Survey on Computer Vision Technologies HASHIMOTO Yasuhiro

01. Smart Museum

Contact
  • YOSHIOKA Rentaro
Mission
  • Role : Improve visitor experience at the Fukushima Museum
  • Target : Visitor support and exhibition design
  • Value : Improve visitor learning and knowledge acquisition
Purpose
  • Currently, it is difficult to obtain objective data on visitor's behavior in viewing the exhibits that is necessary for evaluation/improvement
  • Also, it is currently difficult to understand the visitor's learning experience so as to provide necessary support
  • To improve the situation, it is desired to be able to objectively grasp visitor's behavior within the exhibit hall
  • Furthermore, it is required to collect visitor's learning experience without hindering their experiences
  • Therefore, a system that objectively measures visitor behavior and supports curators for analysis and interpretation will be developed
Goal
  • Visitor's satisfaction in terms of learning improves.
Scenario
  • Realize objective thorough measurement of visitor behavior
  • Develop devices/applications for visitors that encourages/induces appreciation of exhibits
  • Develop advanced computational methods to analyze and visualize visitor behavior
  • Develop a system for curators that encourages analyzing and interpreting the measured/computed visitor behavior
  • Develop a system to assist curators in designing exhibitions based on the analysis and interpretation
Keywords
  • Knowledge experience, activity sensing devices, image recognition, data visualization, data analysis, human-computer-interaction, human-computer-collaboration
Research Database

02. High-speed autonomous driving on low-friction road surfaces

Contact
  • OKUYAMA Yuichi
Mission
  • Role : Improve availability of autonomous driving on low-friction surfaces
  • Target : Modeling methods for automotive driving environments and automated learning mechanisms for driving
  • Value : Acquisition of automated driving algorithms on low-friction surfaces
Purpose
  • Currently, automated driving on low-friction surfaces is difficult
  • Driving on low-friction surfaces is different from normal driving
  • Development on real vehicles is difficult due to cost and environmental reproducibility
  • Using radio-controlled drifting cars that simulate driving on low-friction surfaces
  • Obtaining driving data and development of automated driving in small-size cars or simulator are desired
  • Development of a simulator for racetracks that mimic low-friction surfaces and radio-controlled drifting cars
Goal
  • Realization of automated driving of drifting radio-controlled cars on real-tracks
Scenario
  • Development of a simple 3D modeling method for radio-controlled circuits
  • Development of a simulator that simulates the physical behavior of a drift RC car
  • Development of a system to automatically learn to drive on the simulator
  • Development of a system to absorb differences between simulator and real world observations and driving
Keywords
  • 3D modeling, artificial intelligence, reinforcement learning, sim2real, autonomous driving, drifting/sliding
Research Database

03. Smart Learning

Contact
  • WATANOBE Yutaka
Mission
  • Role : Reduce educational disparities in accordance with SDGs 4 (Quality Education for All)
  • Target : Learning support and educational support in the educational field
  • Value : Visualization to motivate learning, explanation methods to facilitate understanding, user interfaces and machine learning models for autonomous learning environments
Purpose
  • There are educational disparities in education, especially ICT education, due to regional, school, and economic conditions
  • Currently, business and educational methods related to enriching content, training systems, and competitive/gaming elements are the mainstream
  • However, for skills exercises such as programming, for example, easy-to-understand explanations and feedback are necessary, but a lack of instructors is becoming a problem
  • Therefore, we will develop a smart learning environment and its subsystems to support learners' self-directed learning
Goal
  • Improving learner motivation or self-directed learning efficiency
Scenario
  • Development of visualization methods to view and manage the state of the learner
  • Development of representation techniques to explain and execute algorithms and procedures
  • Development of user interfaces that can adapt to the learner's situation
  • Development of machine learning models to support autonomous learning
  • Development of data mining and data analysis methods for learning data
Keywords
  • Educational technologies, visualization, educational data mining, adaptive learning, autonomous learning, programming, user interface/experience, artificial intelligence

Research Database

04. Smart Information Visualization

Contact
  • TAKAHASHI Shigeo
Mission
  • Role : Enhanced readability in abstract data visualization
  • Target : Visual metaphor design for interactive visual data analysis
  • Value : Egocentric visual data analysis with Human-in-the-Loop
Purpose
  • The amount of available data continues to grow regardless of its type
  • The demand for extracting important features from such large-scale complicated data also increases
  • This leads to the need of developing egocentric visualization of such extracted data for individual analysts
  • We tackle this challenge by implementing an interactive system for information visualization
  • We also design new visual metaphors to improve the readability of the data
Goal
  • Help analysts understand the data according to individual requirements interactively through visualization
Scenario
  • Prepare the data structure of networks having vertices with multivariate vectors and edges connecting them
  • Find 2D optimal layouts of the networks by referring to inherent relationships and similarities among vertices
  • Transform the networks into visual metaphors by successively deleting minor edges according to their importance
  • Transform the network into visual metaphors by extracting the major connectivity among vertices
  • Visualize the networks from their global structures to local features via visual metaphors by continuously controlling the connectivity among vertices
Keywords
  • Network visualization, network drawing, visual metaphor design, interactive visualization
Research Database

05. Analytical Tools for Network Science

Contact
  • HASHIMOTO Yasuhiro
Mission
  • Role : Development of analytical tools for network science
  • Target : Research and data analysis work on any natural and artificial systems that can be represented as a network
  • Value : Application of network science to familiar data
Purpose
  • Highly developed systems such as the Web, living organisms, and social systems have network structures in which numerous components interact in complex ways
  • Network systems that produce such non-uniform and complex dynamics are called "Complex Networks" and form a new field of research
  • This new way of understanding systems is important for the proper control and development of today's diverse and large complex systems
  • Therefore, we will review the research achievements of the last 20 years in complex network science and familiarize ourselves with the major analytical methods
  • Further, we will develop analytical tools for network science that can be applied to familiar data without detailed knowledge
Goal
  • Implementation of analytical tools for network science, with a theoretical background
Scenario
  • Reading the book "Network Science" (Barabási Albert-László) as an entry point, survey a part of significant papers related to complex networks
  • Learn how to calculate the basic statistics of a network (mean path length, degree distribution, and so on)
  • Learn how to analyze meso-scale structure of networks such as community structure and core-periphery structure
  • Learn mathematical models to generate complex networks
  • Learn how to visualize complex networks
  • Develop network analysis tools based on the platform "Observable", a JavaScript based computational essay tool
Keywords
  • Complex networks, network statistics, evolving networks, dynamics on networks, community structure, core-periphery structure, visualization, Observable (web service)
Research Database

06. Analysis of Infodemic on Social Media

Contact
  • HASHIMOTO Yasuhiro
Mission
  • Role : Elucidate the structure of infodemics on social media and understand its mechanism
  • Target : Daily communication and information gathering using social media
  • Value : Voluntarily and systematically discourage behaviors that contribute to infodemics and prevent polarization and radicalization in the formation of public opinions
Purpose
  • An infodemic (the rapid spread of misinformation, disinformation, and biased information) on social media and the resulting social division has become a big issue in our society
  • Designing a system that mitigates the social division caused by infodemics, echo chambers, and filter bubbles and supports appropriate decision-making by the general public is an important challenge
  • This requires understanding not only the reliability of information sources, but also the structure of information spreading and the recursive reactions of the crowds that are exposed to such information
  • Therefore, we will analyze actual data on several topics discussed in social media to reveal the structure of information spreading
  • Furthermore, we will visualize the findings from a data journalistic perspective
Goal
  • Develop skills to analyze, summarize, and express the qualitative and quantitative characteristics of complex communication on social media
Scenario
  • Learn how to collect data and manage them in a database
  • Learn how to extract words from sentences and perform sentiment analysis and distributed representation
  • Learn how to construct social networks that connect users based on dynamic and static relationships in social media such as retweets, replies, and follows
  • Extract relevant information for topics discussed in social and quantitatively evaluate the overall picture
  • Evaluate the meta-structure of communication using advanced methods such as network clustering
  • Based on data analysis, summarize the major factors that cause infodemics and the resulting social division, into a concise and convincing presentation using modern visualization and web technologies
Keywords
  • Social media, Twitter, infodemic, polarization, echo chamber, filter bubble, fake news, data journalism
Research Database

07. Extensive Survey on Computer Vision Technologies

Contact
  • HASHIMOTO Yasuhiro
Mission
  • Role : Overview the achievements to date in computer vision technology
  • Target : Research and development related to computer vision
  • Value : Current status and future prospects for computer vision research
Purpose
  • Computer vision is a technology that gives computers the ability to perform visual information processing similar to or even better than that by humans
  • With recent advances in machine learning technology, computer vision technology is playing an important role in modern technology, as its range of applications has dramatically expanded to include face recognition, anomaly detection, automatic driving, VR/AR, and so on
  • However, the field is vast and ever-evolving, and it is not easy to propose new research themes based on catching up a number of existing research or to position the vast amount of previous research
  • Therefore, we select a specific field of interest from the broad field generally referred to as computer vision and conduct a comprehensive survey of research papers in that field
  • We will then summarize the current status of research in the field of interest and provide an outlook on future prospects
Goal
  • Keep up with the state-of-the-art practices and propose a new research theme in specific field related to computer vision
Scenario
  • Select a primary area to survey from computer vision, such as XR, VSLAM, object detection/tracking, and make a list of research papers to cover in the survey
  • Determine a format for the survey and periodically review and present a set number of papers
  • The survey will be divided into three categories: those papers that are only overviews in order to position them within the field, those that are technically interesting and reviewed in-depth, and those that are important and should be read closely
  • If we find new papers to cover from the papers we have looked through, we will add them to the list as appropriate
  • If there are available implementations for the surveyed papers, we will actively trial them
  • Finally, summarize the survey and its prospects. The target number of papers to be surveyed is approximately 100 during a semester
Keywords
  • Computer vision, virtual reality (VR), augmented reality (AR), visual simultaneous localization and mapping (VSLAM), image processing, image classification, object detection, object tracking, pose detection
Research Database