SIMULATING DISTRIBUTED ALGORITHMS FOR LATTICE AGENTS

O. Aichholzer, T. Hackl, V. Sacristán, B. Vogtenhuber, and R. Wallner

We present a practical Java tool for simulating synchronized distributed algorithms on sets of 2- and 3-dimensional square/cubic lattice-based agents. This AgentSystem assumes that each agent is capable to change its position in the lattice and that neighboring agents can attach to and detach from each other. In addition, it assumes that each module has some constant size memory and computation capability, and can send/receive constant size messages to/from its neighbors. The AgentSystem allows the user to define sets of agents and sets of rules, which are applied to all agents. It simulates the synchronized execution of the set of rules by all the modules, and can keep track of all actions made by the modules at each step, supporting consistency warnings and error checking.

AVAILABLE DOCUMENTS

1. O. Aichholzer, T. Hackl, V. Sacristán, B. Vogtenhuber, and R. Wallner.
   Simulating distributed algorithms for lattice agents.
   To appear in Proc. XV Spanish Meeting on Computational Geometry, June 26-28, 2013.
   PDF
2. R. Wallner.
   A System of Autonomously Self-Reconfigurable Agents.
   Diploma Thesis, Institute for Software Technology, Graz University of Technology, 2009.
   PDF.

THE 2-DIMENSIONAL SQUARE-LATTICE SIMULATOR

• Requirements
  The program is developed in Java 1.4.2.15. For running the application the Java JRE 1.4 or higher must be installed.
• Instructions for installing the simulator
  Download and unzip the file AgentSystem2D_JAR.zip in a folder of your choice. The program folder contains the program file AgentSystem2D.jar and a folder with the files
  • files/config.txt
  • files/agents.txt
  • files/rules.txt
  The file files/config.txt is necessary to run the program. The files files/agents.txt and files/rules.txt are used as the starting initial setting if they are in the folder files.
• The source code
  The source code of the simulator is available. You are allowed to modify it, as long as the following requirements are fulfilled:
  1. The source code of your modified version of the simulator must be made public.
  2. Acknowledgment to the original version and authors of the simulator must be included.
  3. A notice must be sent by e-mail to the authors of the original simulator, explaining what your modification does and where it is available.
• The user's guide
  Here is a user's guide of the system.

2-DIMENSIONAL EXAMPLES OF AGENTS AND RULES

• Self-organizing tasks
1. Compute a spanning tree, if a leader is known.
2. Elect a leader, if the number of agents is known.
3. Count the number of agents, if a spanning tree is available.
4. Solve the three previous tasks (computing a spanning tree, electing a leader, and counting the number of agents) from scratch.
5. Find a balanced position for a root in a spanning tree.
6. Elect a leader by boundary traversal.
7. Compute the minimum bounding box, if the number of agents is known.
• Path finding tasks
1. Escape a maze.
2. Enter, traverse, and escape a maze.
• Reconfiguration tasks
1. Sort agents, if id's are known.
2. Reconfigure a strip into a rectangle.
3. From arbitrary shape to histogram.
4. From current histogram to goal histogram.
5. From histogram to goal shape.
6. Reconfigure between two arbitrary shapes (puts together the previous tasks 3, 4 and 5).

THE 3-DIMENSIONAL CUBIC-LATTICE SIMULATOR

• Requirements
  The program is developed in Java 1.4.2.15. For running the application the Java JRE 1.4 or higher must be installed. The 3D system additionally uses the Java3D package (version 1.4.0.01).
• Instructions for installing the simulator
  • Linux
    Download and unpack the tar file AgentSystem3D_Linux.zip in a folder of your choice. The unzipped folder AgentSystem3D_Linux contains the following installer files:
    • install_agentsystem3D
    • install_java3D
    • install_all
    The script install_agentsystem3D unzips the AgentSystem3D to the folder AgentSystem3D. The script install_java3D unzips the java3D package to the JAVA HOME directory (write permissions to install Java3D are necessary). The script install_all unzips the AgentSystem3D and the Java3D package.
    The program folder contains the program file AgentSystem2D.jar and a folder with the files
    • files/config.txt
    • files/agents.txt
    • files/rules.txt
    The file files/config.txt is neccessary to run the program. The files files/agents.txt and files/rules.txt are used as the starting initial setting if they are in the folder files.
  • Windows
    Download and unzip the file AgentSystem3D_Windows.zip in a folder of your choice. The unzipped folder AgentSystem3D_Windows contains the following installer files:
    • Install_Agentsystem3D.bat
    • Install_java3D.bat
    • Install_all.bat
    The script Install_Agentsystem3D.bat unzips the AgentSystem3D to the folder AgentSystem3D. The script Install_java3D.bat unzips the Java3D package to the JAVA_HOME directory (write permissions to install Java3D are necessary). The script Install_all.bat unzips the AgentSystem3D and the Java3D package.
    The program folder contains the program file AgentSystem3D.jar and a folder with the files
    • files/config.txt
    • files/agents.txt
    • files/rules.txt
    The file files/config.txt is necessary to run the program. The files files/agents.txt and files/rules.txt are used as the starting initial setting if they are in the folder files.
• The source code
  The source code of the simulator is available. You are allowed to modify it, as long as the following requirements are fulfilled:
  1. The source code of your modified version of the simulator must be made public.
  2. Acknowledgment to the original version and authors of the simulator must be included.
  3. A notice must be sent by e-mail to the authors of the original simulator, explaining what your modification does and where it is available.
• The user's guide
  Here is a user's guide of the system.

3-DIMENSIONAL EXAMPLES OF AGENTS AND RULES

• Self-organizing tasks
1. Compute a spanning tree, if a leader is known.
2. Elect a leader, if the number of agents is known.
3. Count the number of agents, if a spanning tree is available.
4. Solve the three previous tasks (computing a spanning tree, electing a leader, and counting the number of agents) from scratch.
5. Find a balanced position for a root in a spanning tree.
6. Compute the minimum bounding box, if the number of agents is known.
• Reconfiguration tasks
1. Reconfigure a rectangle into a cube.

RELATED WORK

• Generic distributed universal constant-force reconfiguration of 2D lattice-based modular robotic systems, by Ferran Hurtado, Enrique Molina, Suneeta Ramaswami and Vera Sacristán. Link.
• Fast in-place reconfiguration of expand-contract modular robots, by Manuel Perera and Vera Sacristán. Link.