Location : SYMBRION REPLICATOR >
Print

Simulator


Table of contents


The software simulator is one of the important parts of the projects; it allows off-line and offline-to-online simulations of different evolutionary, self-organizing and other approaches.



Robot3D

The simulator has been made open-source and can be found on Launchpad from which you can access code (using the Bazaar version control system) and Ubuntu (10.04, 10.10) packages.


Background information


It is based on the Delta3D open source gaming & simulation engine. Based on this engine, we are developing Symbricator3D, a simulation tool especially for modular and swarm robots (L. Winkler and H. Wörn. Symbricator3D - A Distributed Simulation Environment for Modular Robots. In Proc. of the 2nd International Conference on Intelligent Robotics and Applications (ICIRA-09), Lecture Notes in Computer Science, Vol. 5928, Springer Berlin / Heidelberg, pp. 1266--1277, ISBN: 978-3-642-10816-7, doi: 10.1007/978-3-642-10817-4\_127). Symbricator3D is written in C++ and consists of three main modules, robot actors, robot controllers and simulation components. The Delta-3D GameManager is handling all these modules and is responsible for the communication between them. While the robot actors are the physical and geometrical representation of the robot, the controller class represents the software that will run on the real robot as well. Simulation components are responsible for the control of the simulation, e.g. there can be components for user interaction, components for supporting artificial evolution and for setting up various experiments or for simulating the environment.



The particular tasks a simulator for modular robots has to cover are:

  • Single Robot Control. Each single robot needs to is an entity and has to be simulated separately to help developing swarm algorithms. The single robot can be simulated dynamically, but does not need to be simulated that way. A 2D-simulator is often sufficient enough.
  • Organism Motion. When the robots are connected together, they build an organism which also has to be simulated. As the organism often accomplishes movements in the 3D-space, a 3D-simulator, that either simulates the movements kinematically or dynamically is essential.
  • Evolving Organisms Structures. Like natural evolution, different organism configurations will compete each other in every generation. The best configurations will be used for the next generation to form slightly different organisms. Over time the population improves, and eventually a satisfactory organism will be evolved that can be used on the real robot. The simulator has to provide functions to support organism evolution.
  • Developing and Evaluating Control Algorithms. Like organism structures, organism control algorithms can also be evolved and the simulator needs also provide functionality for evolutionary control algorithms.
  • Creating Test Scenarios towards the Grand Challenges. Grand Challenges are defined to identify the progress of the projects. To evaluate the quality of the software, the simulator has to support the Grand Challenge scenarios and therefore needs to run very stable while simulating hundreds of robots.



The following sections are devoted to different elements of the simulator (not all from them are currently open to public viewing).

Delta3D: Description and Installation



Documentations, How-Tos and Get-Started



Improvement of Symbricator3D




Controllers


This section is limited to registered users.

  1. Application scenarios
  2. Organism Gait Learning Scenarios
  3. Symbricator3D and Sensor Fusion
  4. Collaborative obstacle bypassing scenario of autonomous SYMBRION-REPLICATOR robots
  5. Current functionalities provided for the Controller
  6. Symbricator3D and Sensors
  7. Symbricator3D and TCP-IP
  8. The first controller



Created by admin. Last Modification: Wednesday 03 of November, 2010 09:49:45 CET by anne.

Menu

Home