Setting up SITL using Vagrant¶
This article explains how to set up the SITL ArduPilot Simulator in a virtual machine environment using Vagrant, and connect it to a Ground Control Station running on the host computer. This approach is much easier and faster than manually setting up a virtual machine to run SITL on Mac OSX or Windows (or Linux).
These instructions have been tested on Windows 8.1.
The SITL (Software In The Loop) simulator allows you to run Plane, Copter or Rover without any hardware. The simulator runs the normal ArduPilot code as a native executable on a Linux PC. SITL can also be run within a virtual machine on Windows, Mac OSX or Linux.
Vagrant is a tool for automating setting up and configuring development environments running in virtual machines. While it is possible to manually set up SITL to run in a VM on Windows (or Mac OSX), it is much easier (and more reproducible) to use Vagrant to do this work for you.
Due to the way submodules are currently handled in the build system, it is not possible to have a repository which can be built on both the host and virtual machines. A dedicated repository should be used for running the Vagrant virtual machine.
Git (1.8.x or later) must be installed on the host computer.
Git for Windows (1.9.5) is recommended.
- The current windows px4 toolchain (v14) does not have a
recent enough version of GIT
- You must use the newer
- version for the git submodule init step. After that setp you can use an older version.
Ensure that git is set to leave line endings untouched. Click on your new “Git Shell (or Bash)” Icon (the terminal was installed when you installed git) and type in the following in the Git “MINGW32″ Terminal window:
git config --global core.autocrlf false
- SSH must be installed on the host computer and be added to the system PATH. SSH is installed with GIT, or you can install it independently for your platform.
Set up the Vagrant and the virtual machine¶
Download and install Vagrant for your platform. Windows, OS-X and Linux are supported.
Clone the ArduPilot Github repository anywhere on your PC:
git clone https://github.com/ArduPilot/ardupilot.git cd ardupilot
Start a vagrant instance
Open a command prompt and navigate to any directory in the /ArduPilot/ardupilot/Tools/vagrant/ source tree.
Run the command:
This starts running a VM, based on a Vagrant configuration file in the source tree. All the files in this directory tree will “magically” appear inside the running instance at /vagrant.
The first time you run the vagrant up command it will take some time complete. The command needs to fetch a Vagrant base VM and configure it with the development environment.
Initialise git submodules
The ArduPilot source tree references other repositories as submodules. These must be initialised by working on the virtual machine:
vagrant ssh cd /vagrant git submodule update --init --recursive exit
Start running SITL¶
Enter the following in your vagrant shell to run the Copter simulator. This will first build the code (if it has not previously been built) and then run the simulator:
vagrant ssh -c "sim_vehicle.py -j 2"
Once the simulation is running, you will start getting information from the MAVLink prompt about vehicle state. For example:
GPS lock at 0 meters APM: PreArm: RC not calibrated APM: Copter V3.3-dev (999710d0) APM: Frame: QUAD APM: PreArm: RC not calibrated
The Copter Simulator is built by default, but you can instead build for
the plane or rover using the
vagrant ssh -c "sim_vehicle.py -j 2 -v ArduPlane" vagrant ssh -c "sim_vehicle.py -j 2 -v APMrover2"
Run Mission Planner or MAVProxy in your main OS¶
You can now connect to the running simulator from your main OS. Just connect to UDP port 14550, either from Mission Planner or MAVProxy. The MAVProxy command is:
Shutting down the simulator¶
When you are done with the simulator:
Press ctrl-d in the Vagrant SSH window to exit the special MAVProxy that is gluing everything together.
Suspend the running VM by entering the following in the command prompt:
Restarting the simulator¶
When you need the simulator again you can resume the VM and restart the simulator as shown:
vagrant up vagrant ssh -c "sim_vehicle.py -j 2"
Restarting the environment usually only takes a few seconds as the VM is only suspended and the simulation code for the vehicle has already been built.
Updating the simulator¶
The simulator is built from the source tree shared between the host and virtual machines, and any changes will trigger a rebuild next time you start the simulator. To update the simulator you simply need to modify the source tree (or pull a new version from Github).