Integrating spatio-temporal data in agent-based simulations for emergency navigation support Zhiyong Wang In cooperation with Sisi Zlatanova 1 Content Background Research question Research methodology The prototype system architecture Use cases Data and software Proposed timetable Planned Publications 2 Disaster management and Emergency navigation Pros and cons of GIS tools Pros database management spatial analysis geographical visualization ………. Cons dynamic modelling simulation 4 Agent based Modeling of Simulation Agent is: - a piece of code to describe dynamic phenomena (moving people, events, plume...) An agent-based model (ABM) is: - a class of computational models - provides simulation of actions and interactions of autonomous agents (both individual or groups) The goal is assessing effects on the dynamic system as a whole (e.g. disaster management, evacuation, natural phenomena, etc. ) 5 Examples of Agent-based modeling Predator Prey model http://www.xjtek.com/anylogic/de mo_models/11/ Swarm following the leader http://www.youtube.com/watch?v=f510pma hpE8 6 Pros and cons of Agent Model simulation Pros Complex dynamics Behavior and interaction between agents Prediction ………. Cons Data used are mostly simulated(no real data) Predetermined model 7 Integration of GIS and Agent-Based Model Simulation Agents and Networks: Commuting Arena Evacuation Simulation in EPT http://www.youtube.com/watch?v=mvkz1 HwEWXU&feature=socblog_th http://www.youtube.com/watch?v=ixTi uLwlLSc&feature=related Deficiencies of the existing integration of ABM and GIS Uses lacks real data Uses real maps and 3D models but hardly real-time data ( e.g. GPS tracks, plume movement) Rigid input parameters Predefined model (No correction) 9 Dynamic Data Driven Application System (DDDAS) DDDAS formulates simulation models and methods with: - dynamically measured data, - algorithms, - system tools, - mathematical and statistical advances The challenges are: - how to incorporate additional data -how to dynamically steer the measurement process? 10 Main Research question Is integration of spatial data (static such as 2D/3D models and dynamic such as real time measurements) into agent-based simulations able to better support the emergency navigation considering human movements and moving disasters? 11 The prototype system architecture 12 1. Research questions related to GIS What kind of information will be needed for agent-based simulation? How to derive the network from these information What kind of relationships between agents and spatio-temporal objects will support the simulation? What kind of data model should be used for management of the dynamic data that includes the real-time data, the information of moving objects (disasters, pedestrian, vehicles, etc.) 13 2. Research questions related to Agents How many types of agents should be developed? What kind of user profile should be considered for route determination under emergencies? What kind of rules should be designed for agents to dictate their behaviors? How can we evaluate the behaviors of the agents(e.g. rescue vehicle agent, etc.)? 14 3. Research questions related to routing What kind of routing algorithms should be considered? What kind of re-routing strategy should be designed for the agent to avoid obstacles and disasters? How does the agent-based simulation system provide navigation services for both first responders and citizens? On what kind of network (2D and 3D) the algorithms will be run? 15 4.Research questions to DDDAS What kinds of real-time information can be incorporated into the simulation model? How to correct the agent-based model with real-time data? How to verify and validate the developed simulation model? 16 Research Methodology Literature study and technology requirements Literature study and technology investigations Conceptual design 1. Design a multi-agent simulation framework for moving objects (considering GPS tracks) 2. Extend the multi-agent simulation network by considering disaster models (considering real measurements) Implementation developing prototype Validation and Adaptation Conduct tests in different scenarios Make improvement based on the test results. 17 Use case 1: Agent-based modeling of human movements during disasters Jam Different pattern s of human movements Flee 18 Demo 1: updating positions with GPS tracks During the demo is running you have to explain in when GPS tracks will be used 19 Use case 2: Agent-based approach for emergence navigation considering the dynamics of disasters 20 Demo 2: moving plume updated with measurements obtained from the field {position, type of geometry, moving speed, direction} 21 Software and Data Software: Programing language: Java Programming tools: Eclipse, Java 3D, Java OpenGL Visualization tool: Bentley Agent-based modelling toolkit: REPAST, MASON DBMS: PostGIS or Oracel Spatial Data: available 2D and 3D data 22 Proposed timetable Year 1 Literature study and technology investigations Create the agents for modeling the movements of rescuers corrected with GPS tracking Make predictions based on mathematical models Implement the data models for moving objects Test the prototype system in the road network Improve the system based on test results Extend the work into 3D environment Proposed timetable Year 2: Design and implement a multi-agent system for emergency navigation Connect the multi-agent simulation model with the disaster model Extend the work into 3D environment Year 3: Test the agent-based simulation system in different scenarios Improve the models of agents. Year 4 Extend the work considering the disasters and communication between agents Assess and improve our systems. Write the thesis. Planned publications Regular reports 2 or 3 conference papers per year e.g. ISCRAM, AAMAS, DDDAS,UDMS etc. 2 papers in reviewed scientific journals e.g. Computational Geosciences, Computers and Geosciences, international journal of emergency response Thanks for your attention! Any questions and suggestions? 26