Skip to content

Digital Rehearsal

We have developedDigital Rehearsal technology that combines AI-based big data analysis with behavioral economics knowledge to accurately reproduce the behavior of people as it changes in response to a situation. Simulating these responses enables us to predict how people’s behaviors may change in response to certain measures and verify their effects and impacts of in advance.

The Behavior Selection Model that Underpins Digital Rehearsal

Fujitsu has developed the Behavior Selection Model , which accurately replicates people’s movements within a Digital Rehearsal.With a unique technology that combines prospect theory, one of the leading theories in behavioral economics, with machine learning, the Behavior Selection Model can express the difference in choices between individuals while maintaining behavior characteristics that are common to many people.

In the Demo App, users can see the results of reacreating people’s movements using the Behavior Selection Model, which chooses the movement method and route with simple screen maneuvers.

Experience Digital Rehearsal technology

  • We have prepared a Digital Rehearsal experience Demo App that enables users to assess the differences in the effects of various measures upon implementation by virtually deploying a shared mobility* virtually in a city
    *Shown as shared e-scooters on the app

Using the Behavior Selection Model in the Demo App

  • In the demo app, the human object registered as human flow data (OD data) moves from the origin to the destination.OD data is a representation of human movement from origin to destination.

  • The Behavior Selection Model selects a mode of transportation using the assessment value (time, cost, etc.) of each method of movement (walking/car/shared e-scooter) from the origin to the destination, and the human object moves based on the selection.

Assessment in the Demo App (GUI)

  • Using the Behavior Selection Model in the Demo App, users can implement changes in the number of scooters docked at a bay or the parameters of measures to experience the different behavior selections that result between individuals and assess the effects of measures.

Use Case: Dynamic Discount

  • Use Case Summary
    • You can measure the effects of the following Purpose1/Purpose2 discount measures.

Purpose 1: Discounts for returning low-battery e-scooters to designated bays

  • Outline of measure
    • Service providers can assess the balance between discount rates that prompt returns of low-battery shared e-scooters to designated bays and the workload for collection for charging.
    • Users can earn discounts by returning low-battery shared e-scooters to designated bays
  • Assess effect of measure using Demo App
    • By making changes to the battery level indicator, which detects low battery levels , and the discount rate, the Behavior Selection Model changes the utilization rate for shared e-scooters
    • At the same time, users can calculate CO2 emissions and revenue differences, which enables the search for the best overall measure

Purpose 2: Discounts for returning e-scooters to bays in short supply

  • Outline of measure
    • Service providers can evaluate discount policies that reduce the number of bays where shared e-scooters are in short supply. In doing so, they aim to avoid lost opportunities and reduce the burden on operators to move shared e-scooters.
    • Users can earn discounts by returning shared e-scooters to bays in short supply
  • Assess effect of measure with Demo App
    • By making changes to the threshold at which a bay is determined to have low scooter levels and the discount rate, users can simulate how many people will make returns to bays in short supply
    • At the same time, users can simulate revenue maximization and CO2 emissions reductions, which enables the search for the best overall measure

Use Case: Road Closure

  • Use Case Summary
    • What if you close a particular road and make cars inaccessible? you can estimate the impact on which mode of transportation people will choose and whether the surrounding roads will not be crowded.
    • It can be used when you want to evaluate the effect of pedestrian mall, road closure due to accidents or road construction.
  • Check the effect of measures with a demo application
    • You can select areas on the map for road closures.
    • By adjusting the size and duration of closed areas, you can compare and evaluate changes in transportation choices and the impact on surrounding roads.

Use Case: Road Pricing

  • Use Case Summary
    • What if you make a certain area a toll area and charge for a car? You can estimate the effect of transportation mode which people will choose and whether the surrounding roads will not be crowded.​
    • It can be used when you want to reduce the car usage and encourage the use of other modes of transportation.​
  • Check the effect of measures with a demo application
    • You can select charging area and price on the map.​
    • By adjusting the charging area, you can assess the impact on surrounding roads and changes in transportation choices.​

Use Case: Park & Ride

  • Use Case Summary
    • How can we set rates for which parking lots to increase public transportation use and effectively reduce CO2? By examining measures to encourage commuters to change their behavior (incentives such as parking fee changes), we will promote park and ride for people commuting from the suburbs to the city center.
  • Check the effect of measures with a demo application
    • You can set up parking lots near stations in the suburbs or specify the parking fee.
    • By adjusting the location and price of parking lots, we can verify the effect of park and ride on reducing car use and reducing CO2 emissions.

Application: Customizing the Demo App

  • Providing registration data enables more customizable simulations

Data Generator Tools

  • It provides map data generation, human flow movement, and station placement changes for e-scooters to make it easier to customize the demo app.

Functions Ovierview
OSM Converter Generate city map data.
Stations Builder Place e-scooter stations data on a map.
OD Builder Register people flow OD data on a map.
model.json Generator Create the setting file for behavior selection model.
precondition.json Builder Create the setting file for the demo app.
Synthetic Stations Generator Generate random location e-scooter station data.
Synthetic OD Generator Generate random people flow OD data.
road-closure.csv road-pricing.csv Generator Generate the setting files for closed and charged roads.

Usage Image: Simulation Results (Preset)

  • With simulation results shared on the top screen, users can easily check the results of the Digital Rehearsal

Initial Data and Configurable Items in the Demo App

  • The initial data and configurable items in the Demo App are as follows.

Classification Item Contents
Model Behavioral Seleciton Model Demo model (Contact us for customization)
Initial Data Map Data Nakahara Ward, Kawasaki City (Todoroki area)
Method of Movement Walking/Car/Shared e-Scooter
Human Flow OD Data Random dummy data
Configurable Item Simulation Duration Modifiable (GUI/File upload)
Number of Bays Modifiable (GUI/File upload)
Location of Bay Modifiable (API/File upload)
Map Data Modifiable (API/File upload)
Human Flow OD Data Modifiable (API/File upload)
Model Behavior selection model Modifiable (GUI/File upload)
Policies Discount for Return of Low-Battery Scooter Selectable (GUI/API)
Discount for Return to Low-Stock Bay Selectable (GUI/API)

List of Provided APIs

  • API for simulation control and result manipulation

Method URL Notes
GET /api/simulations Obtain simulation list
POST /api/simulations Run simulation
GET /api/simulations/<simulationId> Obtain simulation details associated with simulationId
DELETE /api/simulations/<simulationId> Delete simulation associated with simulationId
GET /api/simulations/<simulationId>/<filename> Obtain simulation file associated with simulationId

  • API for data management in simulation (API Server)

Method URL Notes
GET /api/simdata Obtain list of data files for simulation
POST /api/simdataAsForm Upload data files for simulation
PATCH /api/simidata/<simdataId>/metadata Set metadata
GET /api/simdata/<simdataId> Download data files for simulaiton
DELETE /api/simdata/<simdataId> Delete data files for simulation

  • API for calculating metrics from simulation results

Method URL Notes
GET /api/metrics/<simulationId> Obtain index calculation of simulation results

For customers considering further use

Our project is to promote initiatives that take advantage of Social Digital Twin (Digital Rehearsal).Please feel free to reach out to us.

  • Consulation examples
    • I want to do a simulation in my city
    • I want to edit the map data
    • I want support for creating OD data
    • I want to create my own Behavior Selection Model
    • I want to test public transportation (simulations including trains/buses)
    • I want to use it for an actual integration discussion in the share cycle
    • I want to conduct a Digital Rehearsal for something other than Shared Mobility
  • Contact
    • Please contact us via the “Inquiries" button on the Research Portal menu