- Large-Scale Distributed Simulation Platform with a Multi-City-Level Area, Massive Amount of Mixed Domain Agents, and Dynamic Extensibility
- Unified System of Saving and Extracting Spatio-Temporal Data with Different Granularity and Formats
- Generating Virtual People Flow Data Based on Large-Scale Movement History in Urban Areas
Urban Design Group conducts researches on wider area, for more efficient and comfortable life. We are developing systems for storing and extracting spatio-temporal data of different granularity and formats, analyzing human flows using GPS data, and generating pseudo human flows. This group also develops a distributed simulation platform with a multi-city-level area, massive amount of mixed domain agents, and dynamic extensibility.
Large-Scale Distributed Simulation Platform with a Multi-City-Level Area, Massive Amount of Mixed Domain Agents, and Dynamic Extensibility
Simulation of human and traffic flow is used in a variety of fields, including urban planning, evacuation planning, and prediction and analysis of traffic congestion. In this research, we are developing a human traffic flow simulation platform that has four functions: large scale, heterogeneous inter-simulator collaboration, data assimilation, and dynamic extension. A distributed simulation model is employed to process each agent type and area. In addition, a variety of innovations are used to improve performance, such as the ability to distribute to multiple resources and the use of a flexible distributed system, Synerex.
It is becoming possible to obtain a lot of Spatio-Temporal Data due to advances in mobile terminals and sensor technology. However, it has not been sufficiently studied to easily and uniformly handle data from heterogeneous sensors with different sample rates, resolutions, and formats. In this research, we create a system that can convert multiple types of sensor data with different spatio-temporal granularity into the spatio-temporal granularity desired by the user. The database can be divided into a database that can handle granularity-transformable data, a function to store data in a unified format, and a function to extract data at any granularity.
In this research, we generate synthetic human flow data by extracting and modeling the move and stay tendency of users from real travel history data.This data does not contain any personal information like raw GPS data, and it is a whole population dataset.In addition, we aim to create a dataset with higher granularity and more diverse models than the conventional method by improving the human flow modeling method.