Simulation Environment

Simulation Environment (WeBots)

simulation environments provide a physical environment where one can create a model of physical objects and agents (Thymio2 robots). A simulation environment is important to:

• proove desired behavior of robot agents under physical constraints
• proove desired control of the implemented controller algorithm
• acquire knowledge about the system in scenarios that are difficult to provide in reality (many robots, infinite space)
• infere knowledge about the effects of changes in the physical environment and prepare to real experiments accordingly

There are many simulation environments used in robotics on the market, such as Argos, ROS or WeBots. We chose to use WeBots as Thymio2 robot models are already implemented and ready to use here with an arbitrary controller that can be also written in python3.

Installation of VSWRM in WeBots

In this step-by-step guide I show how to install webots and more importantly how to start an example siulation on your system using VisualSwarm robot models.

1. Install WeBots

• Ubuntu

to install webots on an ubuntu system I suggest to use APT as stated in the step-by-step guide on webots' installation page.

• Windows

on a Windows system just download the installer as stated here and follow the installation wizard.

2. Clone VisualSwarm

after installing WeBots choose a folder where you are going to install everything related to VisualSwarm (clone directory, create virtualenv, etc.). Then simply clone the project in this folder.

• Ubuntu

Install git if not yet installed and then use the following commands:

//creating new folder
mkdir Webots-VSWRM
cd Webots-VSWRM

//clone VSWRM
git clone https://github.com/mezdahun/VisualSwarm.git

• Windows

On windows you might need to install git and git bash to be able to use git features.

3. Create a virtual environment

One can create a venv for WeBots according to this very nice description which I will also copy here.

• Ubuntu

  //Installing virtualenv
  sudo apt-get install python3.7-venv

  //Create a new virtualenv for the project
  python3.7 -m venv VSWRMenv

  //Activate the environment
  source VSWRMenv/bin/activate

  //Check if the env is healthy
  which python

after the last command you should see this: .../env/bin/python

• Windows

  //Installing virtualenv
  py -m pip install --user virtualenv

  //Create a new virtualenv for the project
  py -m venv VSWRMenv

  //Activate the environment
  .\VSWRMenv\Scripts\activate

  //Check if the env is healthy
  where python

after the last command you should see this: .../env/bin/python.exe

4. Install VisualSwarm in the virtual environment

after activating the virtual environment we can install the package VisualSwarm in it locally. You can see if the venv is activated as (VSWRMenv) is visible in your terminal in front of the command line.

• Ubuntu and Windows

Please be sure that your venv is active, otherwise you will install the package globally! Here I suppose you are in the directory that you created in the beginning in which you have the folder VisualSwarm that is the result of cloning the project.

Currently the simulation-related changes are on a dedicated branch so first checkout to this branch with

cd VisualSwarm
git checkout feature/webots-simulation

As the simulation and the real robot version of the package is slightly different an additional setup_simulation.py script is is now visible on this new branch. To install the simulation version, first delete the original setup.py and rename setup_simulation.py to setup.py. After this step pip will recognize this setup file and installs the correct simulation version instead of the other one that should only be installed on Raspberry Pis.

As you are about to install the package VisualSwarm on a non-Raspbian system, you need to use the following to install picamera module

export READTHEDOCS=True

Now go back to the parent directory and use the following to then install Visualswarm's simulation version:

cd ..
python -m pip install --upgrade pip
pip install VisualSwarm

If you get any dependency error during installing the package, please visit the installation page on the wiki and follow the step-by-step guide there to resolve these dependency problems.

after installation check if VisualSwarm is installed correctly by using

ENABLE_SIMULATION=1 vswrm-health

for which you should get an output INFO:visualswarm.app:VisualSwarm application OK!

5. Adjust WeBots preferences

Now we need to enable webots to use our created virtualenvironment in which we installed VisualSwarm. To do so, open webots and go to tools>>preferences and set the python command field to the following path: <absolute-path-to-created-virtualenv-parent-directory>/VSWRMenv/bin/python3.7.

You can also set the startup mode field to Pause so that the simulations do not start automatically when webots open.

6. Start webots and run the example project

After configuring the python command now you can open the example .wbt file in the cloned directory, by clicking on File>>Open World... in webots. Browse into the folder where you cloned Visualswarm and see the wbt file under VisualSwarm/data/webots/VSWRM_WeBots_Project/worlds.

You should be able to run this world file after which the robots start moving according to the implemented controller.

The interface code of the controller is in a python file under VisualSwarm/data/controllers/VSWRM-controller. In this folder there are also configuration json files that allow the user or any other software to start the simulations with an arbitrary set of parameters.

More information about these parameters are provided in the controller code and will be further described on another wiki page soon.