Multi-Agent Simulation Framework Development Using Frogs

Multi-Agent Simulation Framework Development
Using Frogs as a Case Study
Søren Vissing Jørgensen, [email protected]
Center for Bio-Robotics, Maersk Mckinney Moeller Institute
RANA, Reproduction of Artificial and Natural Agents
A new multi-agent simulation framework with the following features
High precision broadcast and synchronization of agent actions.
High volume (>100 agents).
Flexible multi-level agent design via the Lua scripting language.
Highly modular design, with cross-platform support.
Multi-thread optimization’s to support modern computing platforms.
Data visualization functionality to enable post-processing of simulation data.
Over Time Visualization
Real-Time Visualization
Support for dynamic environment simulation that include factors such as temperature and humidity.
Build environments based on loading of real maps.
High Performance Natural Agent Data Acquisition Hardware
Development and utilization of recording hardware
Perform sound recording of Natterjack frogs during mating season.
Uncover population density, individual call activity and chorusing precision.
Perform simulations that correspond to collected data.
Uncover needed parameters to achieve agent behaviour that correspond to the Natterjack and other species.
Breaking Down the Barrier Between Artificial- and Natural Agents
Integration of artificial agents into a real environment
Enable the Android platform to load agent behaviours designed in Lua to create artificial frogs.
The Android frog agents will then be placed in a real environment and attempt to integrate with the natural frogs.
Perform tests on whether or not a designed agent is successful in a real environment.
Can be used as a tool to test how local frogs will respond to different behaviours tested.
Project Supervisors:
Project Period:
John Hallam, Yves Demazeau and Jakob Christensen-Dalsgaard
November-2013 to September-2016