-Tokyo Digital Twin Project 2021- Demonstration 01 Report

Transcript

1. Copyright © Mitsubishi Research Institute
   Tokyo Digital Twin Project
   Demonstration 01 Real-time human flow
   visualization including underground space
   Report
   

   View Slide

2. Contents
   1
   1. Background and Overview
   2. Area
   3. Part A: Human Flow Data Collection and
   Visualization
   4. Part B: Disaster Information Provision
   5. Operation
   6. User Survey
   7. Results and Issues
   8. Future Direction
   

   View Slide

3. １．Background and Overview
   2
   

   View Slide

4. 3
   Background
   Information(congestion level, evacuation)
   provision based on real-time human flow data
   ◼ Mechanism to enhance QOL by using outdoor spaces in urban center area
   ◼ Information providing system for people with difficulty returning home in the event of disaster
   ◼ New lifestyle with/after Corona : avoiding closed, crowded, and close-contact settings
   Provision of real-time human flow data from above and
   below ground during daily life and disaster is necessary.
   Evacuation route guidance based on human flow in urban area
   Effective information provision in the event of a disaster
   

   View Slide

5. Goal and Overview
   Overview
   ◼ Route simulation was conducted by merging congestion information and spatial
   information. Route guidance in normal situations was distributed via a web
   application and its effectiveness was verified. (Part A)
   ◼ Route guidance to evacuation sites and disaster information were distributed
   under the assumption of a disaster situation and its effectiveness was verified.
   (Part B)
   ◼ The web application and 3D viewer data was made public, and opinions about its
   effectiveness and problems were collected through a questionnaire survey to
   identify issues for social implementation. (Part A / B)
   Goal
   4
   ◼ Realizing safe and secure life in Tokyo using real-time human flow
   forecasting data
   ◼ Encouraging people to avoid congestion and improve awareness of
   disaster prevention and evacuation
   

   View Slide

6. ２．Area
   5
   

   View Slide

7. 6
   Area: Aboveground (Part A / B)
   出所：Tokyo Digital Twin Smooth NAVI
   【Legend】
   ：Demo Area
   ：Info Device
   Reasons
   Otemachi, Marunouchi, and Yurakucho areas
   were selected as aboveground demo area.
   Many visitors made regular real-time data collection
   easy
   • Largest office district in Japan with large daytime population, made
   it easy to measure human flow.
   • 42,000 people had difficulty returning home on weekdays, while the
   capacity is 21,000 (estimated March 2020), recognized seriously.
   Public-private partnerships made it easy to collect
   and link data
   • Leading area for the Smart Tokyo Implementation Strategy
   • Active Private area management organizations
   • Leading model project for the 2019 Smart City Model Project
   (publicly solicited by the MLIT)
   Existing services could be linked and compared with
   • The issues to be addressed in this demonstration could be clarified
   by comparing with existing services.
   • This could serve as a trial for public-private partnerships in the
   future, such as route search service collaboration and background
   data sharing.
   

   View Slide

8. 7
   Area: Underground (Part A / B)
   The corridor between Shin-Marunouchi Bldg. and Marunouchi
   Bldg. was selected as underground demo area.
   Reasons
   Source: Marunouchi Map（Mitsubishi Estate）
   https://www.marunouchi.com/files/pdf/j_jp_02.pdf
   Vaccin
   ation
   site
   ①
   ②
   ③
   Many passers-by made regular real-time data
   collection easy
   • This corridor is regularly used by commuters and shoppers
   to the Shin-Maru Building and Maru Building, so many
   human flows can be measured.
   Gyoko-dori Street was used as a vaccination site
   • During the demonstration period, Gyoko-dori Street was used
   as a vaccination site, and an increase in human flow is
   expected as a detour route.
   Other conditions
   • Power supply for sensors was available
   • 3D digital map of the area had already been developed by
   the Urban Development Bureau.
   Marunouchi
   Building
   Shin-Marunouchi
   Building
   

   View Slide

9. ３. Part A:
   Human Flow Data Collection and
   Visualization
   8
   

   View Slide

10. Data Collection
    9
    Comparing the characteristics of each sensor,
    smartphone GPS was employed.
    Smartphone GPS Camera Wi-Fi packet sensor Bluetooth
    System
    GPS location data of
    smartphones is
    anonymized to get
    human flow in the
    specific area.
    Optical cameras are
    installed on sidewalks,
    etc., and human flow is
    measured using image
    recognition technology.
    Human flow is
    measured based on the
    number of devices using
    Wi-Fi function in
    smartphones.
    Human flow is
    measured based on the
    number of devices using
    Bluetooth function in
    smartphones.
    Problem
    Entire human flow
    cannot be measured
    due to the dependance
    on certain smartphone
    carriers.
    Privacy issues must be
    considered when facial
    images are acquired.
    Human flow cannot be
    measured when Wi-Fi
    function is turned off.
    Human flow cannot be
    measured when
    Bluetooth function is
    turned off.
    Decision
    Adopted with
    calibration based on
    entire human flow
    measured with other
    devices.
    Not adopted
    due to privacy issues
    Not adopted
    because the ON/OFF
    ratio of Wi-Fi function
    cannot be estimated.
    Not adopted
    because the ON/OFF
    ratio of Bluetooth
    function cannot be
    estimated.
    

    View Slide

11. Forecast data at present and 1 hour later was made based on
    real-time human flow data, calibrated with fixed sensors.
    10
    Aboveground: Real-time Human Flow Data
    Detail
    Data name LocationMind xPop
    About data Data based on GPS data in NTT docomo devices
    Processing
    method
    Human flow in Otemachi / Marunouchi / Yurakucho Area（Daimaruyu area) was analyzed and index
    data of density of human flow was made for each road link. (When route guidance is set as an
    output, showing human flow by road links is more likely to reflect actual congestion on a route than
    showing by groups of buildings. Also, GPS data inside buildings is often missing because they do not
    move, which poses a challenge to accuracy.)
    Note To minimize the influence of time lag, forecasts were made based on current data.
    Decision
    ◼ Forecast data at present and 1 hour later was made based on real-time data provided with a
    delay of about 1 hour.
    ◼ Forecast was calibrated with human flow data collected by fixed sensors (LiDAR, camera).
    ◼ Fixed human flow sensors were installed with permission from building management companies.
    

    View Slide

12. Aboveground: Real-time Human Flow Data
    Fixed human flow sensors were installed in order to count
    all passers-by and improve the accuracy of data.
    Method Environment
    Equipment: Cameras and LiDAR on
    sidewalks, etc.
    A pair of camera and LiDAR measure
    pedestrians on both sides of sidewalks.
    Location: Sidewalks of Naka-dori Avenue, etc.
    Measurement: Pedestrians on Naka-dori Avenue
    11
    

    View Slide

13. Aboveground: Real-time Human Flow Data
    Road network (DRM Link) was used
    in order to visualize congestion level on the roads.
    12
    Road network Mesh Dot Arrow
    Tying the measured
    people's locations to links
    on the road network
    Tying locations of people
    in one mesh
    Displaying people
    measured with a dot or
    other mark
    Showing numbers and
    directions of people by
    arrow size and color
    Source：MLIT PLATEAU （https://www.mlit.go.jp/plateau/）
    ◼ Road network display is suitable to orthogonally arranged roads in Daimaruyu area.
    ◼ Road network display is popular in car navigation system, easy to familiarize with.
    

    View Slide

14. Aboveground: Visualization of Congestion
    3 congestion levels were defined
    and tied to the road links with Map-Matching Method.
    13
    People near a link between intersections
    were tied to the link.
    (using Map-Matching Method)
    Method of tying
    111 112
    Four people are on
    111-112 link.
    111 112
    Congestion level is
    calculated and
    provided.
    Definition of 3 congestion levels
    3 levels were defined based on trial measurements
    of human flow data before the start of the service.
    Congestion
    Level
    Very Crowded Crowded Vacant
    Color Red Yellow Green
    Number of
    people
    Over 3,000 1,500-3,000 under1,500
    

    View Slide

15. Underground: Real-time Human Flow Data
    The number of passers-by was estimated by the sensors,
    with smartphone device information anonymized.
    14
    Detail
    Data name unerry
    About data Radio waves transmitted by smartphones are used.
    Processing
    method
    Index data of density of human flow was generated respectively for each passage in Maru
    Building and Shin-Maru Building, located in Daimaruyu area.
    Note The actual number was estimated based on the number of people passing near the sensors.
    Decision
    ◼ Congestion level forecast at 1 hour later was calculated within the sensors based on real-
    time data provided with a delay of about 1 hour.
    ◼ Human flow sensors were installed with permission from building management
    companies.
    

    View Slide

16. Underground: Real-time Human Flow Data
    15
    Environment
    Underground measurement sensors
    Radiowaves of smartphones were used instead of GPS, and
    the number of passers-by was estimated using IoT sensors.
    Service Temperature
    Interface
    Battery
    Weight
    Size
    Service Humidity
    Water Protection IPX5
    20 - 80%RH
    -20-55℃ (with battery)
    USB Type C, DC Jack
    1800mAh (standart)
    body: 66g, battery: 30g
    L100mm x W65mm x H14mm
    

    View Slide

17. Underground: Real-time Human Flow Data
    16
    3 sensors were installed in the underground passage.
    Sensor location①
    Sensor location③
    Sensor location②
    出所：Marunouchi Map（Mitsubishi Estate）
    https://www.marunouchi.com/files/pdf/j_jp_02.pdf
    Locations of underground sensors
    2 sensors on
    Maru Building side (①②)
    1 sensor on
    Shin-Maru Building side
    (③)
    Vaccin
    ation
    site
    ①
    ②
    ③
    Marunouchi
    Building
    Shin-Marunouchi
    Building
    

    View Slide

18. Underground: Real-time Human Flow Data
    Mesh was employed in order to
    show locations of people underground.
    17
    Mesh Underground
    passage
    Dot Arrow
    Tying locations of people
    in one mesh
    Tying locations to links
    on the underground road
    network if exist.
    Displaying people
    measured with a dot or
    other mark
    Showing numbers and
    directions of people by
    arrow size and color
    Source：MLIT PLATEAU （https://www.mlit.go.jp/plateau/）
    ◼ Mesh display is suitable for closed underground passages.
    

    View Slide

19. Underground: Visualization of Congestion
    18
    Each passage in Maru Building and Shin-Maru
    Building was defined as one mesh respectively
    and the number of people was counted.
    Count within a mesh Definition of 3 congestion levels
    3 levels were defined based on trial measurements
    of human flow data before the start of the service.
    Congestion
    level
    Very crowded crowded Vacant
    Color Red Yellow Green
    Number of
    people
    Over 3,000 1,500-3,000 under1,500
    １mesh each
    Maru
    building
    Shin-Maru
    building
    3 congestion levels were defined based on the number of
    people in meshes measured by IoT sensors.
    

    View Slide

20. 大丸有エリア地下における背景データ
    The 3D digital map data from the Urban Development
    Bureau was used as the background map.
    19
    Source：Tokyo Digital Twin Project 3D Viewer https://3dview.tokyo-digitaltwin.metro.tokyo.lg.jp/
    [Data] Public indoor spaces (underground) Daimaru area 3D digital map data (FY2021/2022)
    Background Map
    

    View Slide

21. Interface of Web App
    Interface were designed clearly with minimal information.
    20
    Home screen
    Title
    Overview
    Main screen
    External link
    Survey
    Help
    Current location
    Disaster mode
    Underground
    Congestion level
    Settings
    Route search
    Route option
    Language
    Expression of the image of route
    guidance inspired by the curves of the
    symbol mark of Tokyo.
    Visible buttons and simple icons
    for high clarity
    Main screen Route option, Language
    Title and overview Base map and current location Options at the settings icon
    

    View Slide

22. Interface of Web App
    Congestion levels, congestion avoidance route,
    and the shortest route were displayed in one map.
    21
    Route guidance Underground congestion level
    Congestion information and route guidance Congestion levels in the range of sensors
    Start
    Goal
    The area of
    congestion
    level display
    Congestion avoidance
    route and shortest route
    on the base map with
    congestion level
    In the image on the right,
    Solid line: Congestion avoidance
    route
    Dashed line: Shortest route
    (can be changed in settings)
    Congestion level in
    partial areas
    No route guidance
    

    View Slide

23. ４．Part B:
    Disaster Information Provision
    22
    

    View Slide

24. Disaster Information Provision
    Necessary information was selected
    assuming the use in an event of disaster.
    23
    Disaster Evacuation Site in Chiyoda City
    Chiyoda Ward Disaster Evacuation Site Map
    Source：Chiyoda Ward Disaster Evacuation Site Map
    （https://www.city.chiyoda.lg.jp/documents/2093/taihibashoannai_1.pdf）
    Kitanomaru Park The East Garden of
    the Imperial Palace
    Kokyo Gaien
    National Garden
    Hibiya Park
    Sanadabori Ground Sotobori Park
    【Evacuation site】
    A temporary evacuation site used immediately
    after the strike of disaster to avoid danger and
    chaos, and to ensure safety.
    （As of December 2020, 6 evacuation centers.）
    Providing information of
    evacuation routes and sites
    Disaster Evacuation Site
    Kitanomaru
    Park
    The East Garden of
    the Imperial Palace
    Kokyo Gaien
    National Garden
    Tokyo Sta.
    Hibiya Park
    Sanadabori Ground
    Sotobori Park
    

    View Slide

25. Disaster Information Provision
    Methods of providing information
    about evacuation routes and sites were examined.
    24
    Route guidance to evacuation sites Locations of Marunouchi Vision
    The route to evacuation sites was shown
    in the 3D viewer.
    Locations of about 100 monitors in the
    Daimaruyu district were provided.
    （Related news and neighborhood information would
    be on the monitors during disaster）
    Source：Marunouchi Media Link
    （http://marunouchi-media-link.jp/asset/pdf/media_link_marunouchi.pdf）
    Locations of Marunouchi Vision
    Digital twin 3D viewer with route guidance to
    evacuation sites.
    

    View Slide

26. End
    Hibiya park
    Start
    Yurakucho
    st.(west
    gate）
    Disaster Information Provision
    A route guidance from Yurakucho Sta.
    to Hibiya Park was made and shown.
    25
    Example of evacuation route Display on the 3D viewer
    S
    G
    Pedestrian viewpoint
    using the stories
    function
    

    View Slide

27. ５．Operation
    26
    

    View Slide

28. Period
    Part A and B started on September 22, 2021.
    27
    [Part A] Congestion levels based on real-time human flow
    Jul Aug Sep Oct Nov Dec Jun Feb Mar
    Preparation
    Implementation
    Wrap up
    Jul Aug Sep Oct Nov Dec Jun Feb Mar
    Preparation
    Implementation
    Wrap up
    Route guidance for
    Hibiya Park is still
    provided
    [Part B] Locations of evacuation sites *Route guidance for Hibiya Park is still provided
    

    View Slide

29. Call for Participants
    The wide range of participants were invited,
    especially those related to the Daimaruyu area.
    28
    Email for
    Digital Twin and
    Smart city
    stakeholders
    Target Recipient/Media
    General
    The expert commission members
    Companies in the commission
    PLATEAU community
    ICF（Initiative for Co-creating the Future）
    The media
    Visitor
    Town information website
    Marunouchi vision
    Individual dissemination to trustee
    acquaintances
    Worker
    Daimaruyu conference
    Mitsubishi Estate-owned building
    tenant companies
    Email subscribers of OMY Area
    Management Association（Ligare)
    Marunouchi.com Facebook
    Marunouchi vision
    Marunouchi.com Facebook
    Area-targeted
    email for
    Daimaruyu
    stakeholders
    

    View Slide

30. ６． User Survey
    29
    

    View Slide

31. Outline of User Survey
    The questionnaires were collected
    from users of web app and 3D viewer, respectively.
    30
    Period October 13, 2021 - February 25, 2022
    Method
    Web questionnaire
    (separate surveys for web application and 3D viewer)
    Target
    Web app: Web app users
    3D viewer: 3D viewer users
    Number of
    responses
    Web app: 753
    3D viewer: 600
    Outline of the survey
    

    View Slide

32. Questionnaire Webpage
    Visitors were led to the questionnaire webpage
    from the official website or each service.
    31
    Link to
    questionnaire
    from web app
    Link to
    questionnaire
    from 3D viewer
    Link to
    questionnaire
    from the
    official website
    

    View Slide

33. Respondent
    The respondents were recruited from
    the visitors of the webpage, the app users etc.
    32
    ACT5 Member Point App Web Survey Monitor
    Target: Residents of Daimaru-Yuri area, etc.
    Period: November 17 to November 25, 2021
    Number of responses: 119 web apps, 84 3D viewers
    Target: Residents of Tokyo who live and work in the
    Daimaru-Yuri district
    Period: December 10, 2021 - December 14, 2021
    Number of respondents: 418 web apps, 417 3D viewers
    App of “Daimaruyu SDGs ACT5” , acting for
    achieving SDGs in the area, was introduced
    and points were awarded to respondents.
    The Internet research service
    provided by Macromill was used.
    

    View Slide

34. Web App: Questionnaire
    Opinions about the feasibility of congestion avoidance
    and areas for improvement were gathered.
    33
    Main question Check point
    Attribute ◼ Gender, Age, and Relationship to Daimaruyu area Differences in trends by attribute
    Web App ◼ Recognition route
    ◼ Reason for access
    Effective method to be recognized widely
    Functions of high interest
    Congestion
    avoidance
    route
    ◼ Use of the route search, purpose, and scenario
    ◼ Adoption of congestion avoidance route
    ◼ Whether congestion could actually be avoided
    ◼ (If congestion could not be avoided)
    Reason, specific date, time, and location
    ◼ (If the avoidance route was not taken) Reason
    Users of the route search
    Actual users of avoidance routes
    Effectiveness of avoidance routes
    Points of improvement in search method
    Congestion
    level display
    ◼ Clarity of above and below ground display
    ◼ Opinions about congestion level display
    Points of improvement in display method
    Disaster
    information
    ◼ Recognition of evacuation sites and Marunouchi Vision,
    and usefulness of location information
    ◼ Effectiveness of offline functions
    ◼ Opinion about information needed during disasters
    Recognition of existing services
    Effectiveness of offline information
    Points for improvement
    Others
    ◼ Expansion in other areas
    ◼ Recommendation of the web application
    ◼ Opinions about difficulties and additional functions
    Needs for web app in other regions
    Points for improvement in general
    

    View Slide

35. Web App: Survey Results (1/9)
    No significant differences were found
    in gender and among generations.
    34
    Respondent attributes（Gender, Age）
    Male
    41%
    Female
    58%
    NA
    1%
    Gender
    20-34
    32%
    35-49
    36%
    50-
    31%
    Age
    

    View Slide

36. Web App: Survey Results (2/9)
    Many of the respondents were
    the workers at Daimaruyu area.
    35
    Respondent attributes（Relationship with Daimaruyu area）
    Work at/often visit
    for business
    74%
    Never/rarely visit
    2%
    Occasionally visit for
    sightseeing, shopping,
    events, etc.
    10%
    Occasionally visit for
    business
    13%
    Live within walking distance/
    Used to live, well know
    1%
    Main relationship with Daimaruyu Area
    

    View Slide

37. Web App: Survey Results (3/9)
    The respondents were mainly interested in
    “Congestion display” and “congestion avoidance route.”
    36
    Web application recognition route and reasons for access
    Notification from the
    company working for,
    31%
    Tokyo Metropolitan Government official HP/SNS, 11%
    Mail
    magazine,
    11%
    Acquainta
    nces, 12%
    Others, 19%
    0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
    Recognition route
    Congestion
    avoidance route
    search, 35%.
    Congestion level
    display above and
    below ground, 38%.
    Disaster information,
    22%
    Other, 5%
    0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
    Reason for access
    

    View Slide

38. Web App: Survey Results (4/9)
    More than half of those who searched for congestion
    avoidance route realized the route information effective.
    37
    Congestion avoidance route
    Use 36% Don’t Use 64%
    Pass 65% Don’t pass 35%
    Avoid 34% Somewhat avoid 58%
    Did you search for
    congestion
    avoidance routes?
    succeeded somehow succeeded
    Passed Didn’t Pass
    Did you pass the
    route indeed?
    Did you succeed in
    avoiding the
    congestion?
    Searched Didn’t search
    

    View Slide

39. Web App: Survey Results (5/9)
    The key point to be improved in congestion avoidance route
    is the search method.
    38
    Congestion Avoidance Route Search (Web App)
    Key Points of Improvement
    ❶ Showing time required
    ❷ Registering and saving searched routes,
    Searching by lot numbers or keywords
    ❸ Searching for barrier-free routes
    ❹ Translating information (ex. Building name) in
    the base map for the English version
    １
    ４
    ３
    ２
    

    View Slide

40. Web App: Survey Results (6/9)
    The method of congestion level display was satisfactory,
    but there was room for continuous consideration.
    39
    Clarity of congestion level display
    Clear
    23%
    Relatively
    clear
    42%
    Relatively
    unclear
    17%
    Unclear
    8%
    No opinion 10%
    Above-ground
    Clear
    15%
    Relatively
    clear
    36%
    Relatively
    unclear
    24%
    Unclear
    14%
    No opinion 11%
    Underground
    

    View Slide

41. Web App: Survey Results (7/9)
    The key point to be improved in congestion level display
    is the clarity and simplicity.
    40
    Clarity of congestion level display
    １ ２
    ４
    ３
    Key Points of Improvement
    ❶ Explanations for how to use
    (adding examples, etc.)
    ❷ Forecast for the next few days
    ❸ Congestion levels of each section, including
    directions and changing trends, making clear
    they are about the sidewalks
    ❹ Concrete indicator of congestion level
    ➎ Good color design based on the Tokyo Color
    Universal Design Guideline
    ➏ Current location, even in underground spaces
    ➐ Route search including underground routes
    ５ ６ ７
    

    View Slide

42. Web App: Survey Results (8/9)
    Among users of offline disaster information provision,
    about 70% responded it was effective.
    41
    Disaster information provision
    Effective
    74%
    Ineffective
    7%
    Not sure
    19%
    Effectiveness of providing offline information
    in the event of disaster
    

    View Slide

43. Web App: Survey Results (9/9)
    There is a need for real-time information on congestion
    as well as locations of evacuation sites.
    42
    Disaster information provision
    Key Points of Improvement
    ❶ Voice guidance is desirable.
    ❷ The entrances of evacuation sites can be showed.
    ❸ Congestion levels of evacuation sites can be
    showed.
    ❹ Evacuation routes can be displayed automatically .
    ➎ It would be better to show the safe evacuation
    route, congestion levels, distance, time required,
    safe route home, and estimated waiting time if
    unable to return home, etc.
    １
    ２
    ３
    ４
    ５
    

    View Slide

44. 3D Viewer: Questionnaire
    Whether the contents had helped users to image their
    evacuation route was verified.
    43
    Main questions in the 3D viewer survey Points to be checked
    Attribute ◼ Gender, age, and relationship to
    Otemachi/Marunouchi/Yurakucho District
    Different trends made by attributes
    3D Viewer ◼ Reasons for access Functions of high interest
    Congestion
    level display
    ◼ Clarity of above and underground congestion
    displays
    ◼ Opinions about congestion level display
    Points for improvement of congestion level
    display
    Evacuation
    route
    guidance
    ◼ Opinions about how the evacuation route
    guidance is displayed
    ◼ Imaginability of evacuation routes
    ◼ Information/functions needed to simulate
    evacuation routes
    Points for improvement and effectiveness of
    evacuation route guidance
    Other
    ◼ Expansion in other areas
    ◼ 3D Viewer recommendation rate
    ◼ Opinions about desirable features to be added
    or ones found difficult to use
    Development needs in other regions,
    Points for improvement of the 3D viewer in
    general
    

    View Slide

45. 3D Viewer: Survey Results (1/6)
    Over-50-year-old respondents for the 3D viewer were small
    in amount, compared to those for the web app.
    44
    Respondent Attributes (Gender and Age)
    Male
    51%
    Female
    48%
    NA
    1%
    Gender
    20-34
    34%
    35-49
    38%
    Over 50
    28%
    Age
    

    View Slide

46. 3D Viewer: Survey Results (2/6)
    Many of the respondents were workers at Daimaruyu area,
    the main target of the survey.
    45
    Respondent attributes (Relationship with Daimaruyu area)
    Work at/often visit
    for business
    80%
    Never/rarely visit
    2%
    Occasionally visit for
    sightseeing, shopping,
    events, etc.
    9%
    Occasionally visit for
    business
    8%
    Live within walking distance/ Used to
    live, well know.
    1%
    Main relationship with Daimaruyu area
    

    View Slide

47. 3D Viewer: Survey Results (3/6)
    Users of the 3D viewer were highly interested in 3D digital
    maps in general.
    46
    Reasons for accessing the 3D viewer
    The congestion level
    display above and
    underground was
    interesting, 33%
    The evacuation route
    guidance was
    interesting, 23%
    3D digital map in
    general was
    interesting, 41%
    Others, 3%
    0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
    Reasons for access
    

    View Slide

48. 3D Viewer: Survey Results (4/6)
    Above-ground congestion level display was satisfactory, but
    underground display could be improved.
    47
    Clarity of congestion level display
    Clear
    15%
    Relatively
    clear
    27%
    Relatively
    unclear
    22%
    Unclear
    28%
    No opinion
    8%
    Underground
    Clear
    20%
    Relatively
    Clear
    37%
    Relatively
    unclear
    19%
    Unclear
    15%
    No opinion
    9%
    Above-ground
    

    View Slide

49. 3D Viewer: Survey Results (5/6)
    About 70% users of the evacuation route guidance answered
    they were able to image their evacuation routes.
    48
    Effectiveness of evacuation route guidance during disaster
    Had clear
    image
    69%
    Didn’t have clear
    image
    31%
    Image of evacuation route
    

    View Slide

50. 3D Viewer: Survey Results (6/6)
    Additional information is needed to let users have clearer
    image of evacuation routes.
    49
    Evacuation route guidance
    Key Points of Improvement
    ❶ Names of streets and intersections
    ❷ Street trees and landmarks
    ❸ Images and names of buildings
    ❹ Length and time required of routes
    ➎ Suggestions for alternative routes
    ３
    ２
    １
    ４
    ５ For use in the event of a disaster
    ◼ Usable size of data for smartphone
    ◼ Switchable to 2D
    ◼ Real-time information such as dangerous
    locations
    

    View Slide

51. ７．Results and Issues
    50
    

    View Slide

52. Results
    The usefulness was verified of information provision on
    congestion and disaster using real-time human flow data.
    51
    ◼ Real-time human flow forecast data was provided via the web app for Daimaruyu area, including underground
    spaces, where tend to be crowded.
    ◼ Survey results confirmed that providing congestion levels and congestion avoidance routes was effective for
    encouraging congestion avoidance behavior.
    → A certain number of the web app users searched congestion avoidance routes, and about 70%
    of them adopted the searched routes, with about 90% of them realizing its usefulness.
    [Part A] Real-time human flow forecast data was provided and its usefulness was verified.
    ◼ 3D evacuation routes and other practical data were provided via the 3D viewer and the web app for Daimaruyu
    area to improve awareness of disaster prevention in daily life and encourage evacuation in the event of a disaster.
    ◼ Survey results confirmed it was effective to provide services for disaster (3D evacuation route guidance, offline
    information provision during disasters) in combination with daily services.
    → About 70% of the respondents answered the evacuation route display in 3D viewer was useful.
    → About 70% of the respondents answered offline information during disasters was useful.
    [Part B] Evacuation routes were visualized 3D and improvement in awareness was verified.
    

    View Slide

53. Issues (1)
    Clarity and simplicity of the service should be improved.
    52
    ◼ 3 color-coded levels may not be enough to show situations of congestion objectively.
    ◼ More specific information on congestion level is needed to encourage behavior change.
    Display method of congestion level
    ◼ A certain amount of time is required before real-time human flow data displayed on screen due to
    preprocessing processes such as map matching, traffic mode determination, stay object determination,
    and concealment.
    ◼ To be more “real-time”, a logic to make the preprocessing quick and accurate should be developed.
    Accuracy and speed of congestion level display
    ◼ Data volume was not the point in this demonstration, since the purpose was letting users to check
    evacuation routes in advance and foster evacuation awareness in daily life.
    ◼ Supposing a disaster where services cannot be provided continuously with telecommunication
    restricted, the amount of provided data should be revised carefully.
    Data volume for use in the event of a disaster
    

    View Slide

54. Issues (2)
    Displayed congestion levels did not always reflect actual
    situations accurately.
    53
    The system showed this vacant area as “a little crowded”.
    Cause 1: In urban short trips, GPS accuracy was lowered by many tower blocks.
    Cause 2: The prediction method of human flow was not satisfactory.
    Cause 3: Error due to the difference in the time when data was calibrated.
    Data collection
    Processed past data
    Predictive model
    learning
    Preprocessing
    Future forecast
    Calibration and
    anonymization
    Send to Web system
    

    View Slide

55. Issues (3)
    The service should be easy to use for everyone.
    54
    ◼ Color Universal Design
    → In this demonstration, three colors (red, yellow, and green) were used for color-coding
    of congestion levels. The color-coding should comply to the Tokyo Color Universal
    Design Guidelines to be used by various people.
    ◼ Language support
    → The web application basically supported English, but some parts were displayed in
    Japanese, limiting information in English. The same amount of information should be
    provided in other languages.
    ◼ Voice guidance
    → Voice guidance of congestion avoidance route can be helpful for visually impaired users.
    Universal design
    

    View Slide

56. Issues (4)
    It is essential to expand over a wider area with safety and
    security.
    55
    ◼ System for utilizing data from various devices in a wide area like entire Tokyo
    → Multiple entities will install various sensors so the system for integrating data is necessary.
    → A method is needed to utilize GPS data in areas where data is kept secret due to small population.
    Integrated human flow data for a wide area
    ◼ Privacy consideration in the collections of personally identifiable information
    → Even when anonymized, individuals may be identified by combining other data.
    Regulations for data use in compliance with the revised Personal Information Protection Law and
    other related laws and regulations.
    Privacy-conscious collection and use of data
    

    View Slide

57. Information and Functions Needed
    Useful information and functions were identified for
    congestion avoidance and evacuation.
    56
    Information Function
    Daily
    life
    Real-time ◼ Lifeline (road closures, traffic congestion, etc.)
    ◼ Live video (for grasping congestion levels)
    ◼ Congestion level in buildings and stores
    ◼ Congestion level in rest areas (place to sit, etc.)
    Route
    search
    ◼ Search by lot number or
    keyword
    ◼ Underground area
    ◼ Barrier-free route
    Others ◼ Underground exit numbers
    ◼ Barrier-free information (steps, etc.)
    Others ◼ Display of current location
    including underground
    ◼ Display of time required
    Disas-
    ter
    Real-time ◼ Lifeline (public transportation status, etc.)
    ◼ Live video (for grasping damage in surrounding area)
    ◼ Congestion of disaster evacuation sites
    ◼ Acceptance status of each building for people with
    difficulty returning home
    ◼ Congestion and availability of accommodations
    ◼ Operational status and availability of restrooms,
    elevators, free Wi-Fi, rechargeable spots, etc.
    Route
    search
    ◼ Safe routes to evacuation
    shelters or homes
    ◼ Alternative routes
    Others ◼ Entrance of disaster evacuation sites
    ◼ Evacuation shelter (where, how to use)
    ◼ Bases for relief supply distribution (where, how to use)
    ◼ Locations of restrooms, elevators, free Wi-Fi,
    rechargeable spots, etc.
    Others ◼ Display of current location
    ◼ Voice guidance
    ◼ Display of time required
    

    View Slide

58. ８．Future Direction
    57
    

    View Slide

59. 58
    Future Direction
    Technical and operational issues will be addressed and a
    system for expanding over broader area will be considered.
    We aim to realize safe and secure life in Tokyo
    by providing information of congestion and disaster
    based on real-time human flow data.
    Technical issues Operational issues
    ◼ System for expanding over
    broader areas in Tokyo
    ◼ Method of acquiring and
    using human flow data with
    the consideration of privacy
    ◼ Congestion display method
    ◼ Accuracy of congestion
    information
    ◼ Reliability in the event of a
    disaster
    ◼ Universal design
    ×
    

    View Slide