RoboCup-Rescue Simulation Project Overview

Disaster Problem

At 5:47 AM of January 17, 1995, Hanshin-Awaji Earthquake of a moment magnitude 6.9 hit 20 x 1 km area of Kobe City, Japan, directly killing over 6,432 people, and crushed houses for one-fifth of city's 1.5 million people. 530,000 buildings were damaged, and only 20% of them were usable after the earthquake. The cost for basic infrastructure damage exceeded 100 billion US dollars, and total property loss including private properties well exceeded 300 billion dollars. It was at least ten times larger devastation than 1994 Northridge earthquake hit southern California area. Similar tragedies also took place in Turkey, Taiwan, and other places in the globe.

Watching a number of people dying, the incompetence of our technology was imprinted on every neuron in our whole bodies.

IT Challenge in Emergency Response

The following problems of information technology in an emergency were reported by researchers in Hanshin-Awaji Earthquake.

1. Insufficient supposed scale
2. Damage of Emergency Response centers and members
3. Cut and congestion of communication
4. Information isolation of civilians and volunteers
5. Insufficient information support in decision making

RoboCup-Rescue Simulation Project is a challenge to solve these problems especially focusing on

1. Acquisition, accumulation, relay, selection, analysis, summarization, and distribution of necessary information
2. Sufficient decision support
3. Distribution of systems for their reliability and robustness
4. Continuity of operation from ordinary time to emergency

Goal

The final goal of RoboCup-Rescue is to create the safe social system worldwide.

RoboCup-Rescue Simulation Project challenges this serious problem mainly by multi-agent system technology, large-scale simulation, distribution and acquisition of information, human interface such as PDA and wearable computers, comprehensive networked system technology in addition to existing disaster prevention technologies.

The key points of the project goal are as follows.

1. Inter-operability:
   Software, robots, and other equipment that comply with RoboCup-Rescue standard should have a guaranteed level of inter-operability. This ensures that robots in one region of the world can be deployed to save people in other region most effectively.
2. Open-ended System:
   The whole system shall be built open-ended system, so that any new module and technologies can be plug-in easily and the system can be scalable.
3. Best Practice Configuration:
   The system should be the collection of the best available modules and technologies. The selection of each module and technologies shall be selected competitively.

Road Map

Architecture

The RoboCup-Rescue Simulation Project concept is shown in the figure.

1. Distributed Computation and Simulation:
   Distributed architecture enables this system to integrate heterogeneous components (simulators, agents, GIS, sensory systems, controllers, human, robots, etc.).
   Development cost is minimized by rapidly connecting existing software components by network.
2. Comprehensive Disaster Simulation:
   Integration of various simulators (geotechnical simulation, structure destruction, fire spread, logistics, liquefaction, landslide, tsunami wave, etc.) creates realistic virtual disaster field.
3. Agent Behavior of Emergency Response
   Action of autonomous subjects in this disaster field creates a virtual society. Effect of various actions (strategic behavior, panic action, organization hierarchy and order system, logistics, etc.) are simulated.
   Action to minimize the damage is planned and tested in this virtual society. Competition of the best behavior devises practical strategy and tactics useful in disaster.
4. Real World Sensor Interface
   Connection to infrastructural sensory systems (seismic meter, satellite data, helicopter image, etc.) synchronizes the virtual world with the real disaster providing ability of realtime sensing, prediction and action effect investigation.
5. Mission-Critical Man-Machine Interface
   Planned strategic action is distributed to the human in disaster (emergency response center, rescuer in operation, etc.) for their decision support.
   Information collected by human brigades are sent and distributed by the system.
   Realtime consulting is possible.
6. Control of Infrastructural Systems
   Infrastructure for disaster relief (electricity, gas, traffic, rescue robots) are controlled.

Version 0 Simulator

Prototype simulator (version 0) is developed for

1. Clarifying technological/scientific problems
2. Organizing research members as a Linux-type forum

Kernel

Kernel, a key module of distributed simulation, was developed by Tetsuhiko Koto and Ikuo Takeuchi. It controls communication between every modules and maintains state variables of this virtual field.

Simulators

The prototype includes 4 disaster simulators.

Building collapse simulator was programmed by Kenji Tayama and Fumitoshi Matsuno using Hironao Takahashi's model. This model is based on the Hanshin-Awaji Earthquake data of relation of ground surface acceleration, structure and age with destruction level.

Road blockage simulator was also made by Kenji Tayama Fumitoshi Matsuno and Hironao Takahashi. It uses Hanshin-Awaji data about relation of seismic scale and street width with probability of road obstruction.

Fire spread simulator was programmed by Takechi Matsui on the basis of Takai's model in cooperation with Kobe Municipal Government Disaster Center. It makes micro simulation by combustion, propagation, ignition and extinguish process models.

Traffic simulator was developed by Toshiyuki Kaneda and Masayasu Atsumi. It uses a rule-based micro-simulation method of complex systems considering road width, number of lanes, footpath width, traffic signal, left-turn shortcut, right-turn pocket, etc.

Geographic Information System

DiMSIS (Disaster Management Spatial Information System) developed by Michinori Hatayama and Fumitoshi Matsuno was used for GIS providing real data of Nagata Ward, Kobe City. It conforms of RARMIS (Risk-Adaptive Regional Management Spatial Information System) concept proposed by Kameda and KIWI data format of ISO automobile navigation system standard elect.

Agents

Autonomous agents has the following sensing and action abilities.

1. Sensing: see, hear, listen
2. Action: move, say, tell, extinguish, stretch, rescue, load/unload, clear

Using these abilities, the following class of agents can be developed.

1. Moving agents: civilian, fire-fighter, rescuer, police
2. Static agents: fire station, police station, hospital, refuge, etc.

Some sample agent codes created by Tetsuhiko Koto, Masayuki Ohta, Nobuhiro Ito, Takayuki Ito, and Ranjit Nair are open to public.

Michael Bowling developed ADK (Agent Developers Kit) to accelerate the agent programming.

Viewers

A simple viewer is developed by Tetsuhiko Koto. (image)

2D logViewer including intelligent monitoring function is developed by Yoshitaka Kuwata. (image)

3D viewer is being developed by Atsushi Shinjoh and Shigeki Yoshida. (image)

Practical Applications

Some local governments are interested in this version 0 simulator for residential training and education.

The 2000 Nagata Ward Disaster Prevention Practice in Kobe City used the version 0 simulator for the operation guidance. (image)

How can I get involved in?

Participation in this project is welcome from every aspects, for example, in the following ways of cooperative research.

Multi-Agent Researchers:

Effective rescue/fire-fighter/police agents, realistic civilians, organization stations, and everything are welcome.
In the 2001 competition, the intelligence of disaster relief agents are contested and evaluated.

Disaster Simulation Researchers:

Accurate/efficient simulation modules to make comprehensive disaster field is welcome.
If you have your own simulation engine, plug-in with the version 0 simulator is easy. Development of network communication enables the integration.
If you are interested in comprehensive simulation for decision support, the RoboCup-Rescue simulator is a good fundamental tools. If you develop your own module, you will have the other module including GIS and simulation viewers.

Human Interface Researchers:

PDA, wearable interface, computer graphics, etc. are one of the key issues of RoboCup-Rescue. Connection to the version 0 simulator directly enables disaster managers and rescuers to test the performance, residents to collect local information, and disaster volunteers to know the best way of contribution.

Computer Architecture Researchers:

Performance of computation is an important issue in large-scale simulation. Computer software/hardware architecture research of distributed computation is necessary for future practical application

Disaster Researchers:

Public domain software of RoboCup-Rescue can be used for various analyses.

Disaster Managers, Governments and Local Governments:

Cooperation for applying RoboCup-Rescue simulator to your cities/regions is welcome.
Advice aiming at the future practical application is necessary for this project to proceed in the appropriate direction.

Duty of Participants

For Researchers, Academic Faculties and Students

For research use, all the software resources of RoboCup-Rescue is free of charge. However, in your publication, acknowledgment should be written and some papers should be referred. All the publications should be sent to the technical committee, preferably as an electric file.

Funding is an important issue for all the researchers. We are positive in helping your efforts.

For Governments and Local Governments

We are positive in applying our research results to real situation. Discussions with some local governments are proceeding now.

Necessary cost in application should be prepared by the local government.

Members

Contact

Satoshi Tadokoro
Email: tadokoro@cs.kobe-u.ac.jp
Telephone: +81-78-803-6229
Fax: +81-78-803-6390

Major History