A GUI for running experiments using Kilobot Smart Arena with four tracking cameras
Ubuntu is the preferred OS for ARK
KilobotArena is a Qt program, and therefore uses the Qt build tools to generate the make file. There are two ways to do this - in both cases you need a recent version of Qt v.5.6+ installed - on Ubuntu the version in aptitude should do (sudo apt-get install qtcreator should install everything from Qt that you need).
You also need a CUDA supporting version of OpenCV 3 - you'll need to compile this yourself - here's a guide that should work.
Now either use the QtCreator gui you installed in the previous step (click the hammer button to build - it will prompt you to choose), or run qmake in the ARK directory to generate the Makefile. There are lots of online guides to help if you have trouble e.g.
QtCreator will run the Makefile for you - from the command line you'll have to do it yourself. If you get build errors you may need to change the path to OpenCV in the .pro file (the syntax is quite simple).
You also need to install the calibration program to generate calibrated camera maps: KilobotArenaCalibration.
If you use or adapt ARK in order to generate experimental results, please cite the following paper in any resulting publications:
- Reina A., Cope A.J., Nikolaidis E., Marshall J.A.R., Sabo C. (2017) ARK: Augmented reality for Kilobots. IEEE Robotics and Automation Letters 2, 1755-1761.
- ARK makes use of a redesigned overhead controller: ARK_OHC
- Kilobot Wiki
CUDA can be installed from the official repo (online installation is best) just be carefull to call the correct version of cuda-* instead of simply cuda. WARNING: Do not install CUDA 11 since it is not working with opencv 3
Follow Vanders guide https://github.com/vanderfreitas/KilobotArena to install OpenCV3.4.1 (I had issues with 3.4.5 and PvAPI) or follow this guide and be sure to add:
-D BUILD_opencv_cudacodec=OFF -D WITH_PVAPI=ON -D PVAPI_LIBRARY="/usr/local/lib/libPvAPI.a"
when running cmake
We are using a AV Mako camera, then we have to include its support in the OpenCV installation. Go directly to the OpenCV setup this is not your case. We use the AVT Gige SDK for the camera. The new VIMBA SDK replaced the AVT Gige, but the OpenCV version we are using does not support it yet. We followed the steps from this link: https://github.com/sightmachine/SimpleCV/wiki/Allied-Vision-(AVT)-GigE-Camera-Installation-Guide-for-Ubuntu-Linux
Download the SDK "PvAPI SDK for Linux v1.28" from: https://www.alliedvision.com/en/support/software-downloads.html Choose: "Legacy Software (SDKs, apps, adapters, and interfaces)" Unzip it at /opt with writing permission and:
cd AVT\ GigE\ SDK/inc-pc
sudo cp * /usr/local/include/
cd ../lib-pc/x64/4.4 sudo cp * /usr/local/lib/ cd ../../../bin-pc/x64 sudo cp *.so /usr/local/lib/ sudo cp SampleViewer CLIpConfig /usr/bin
Open the /etc/ld.so.conf file:
sudo nano /etc/ld.so.conf
Add the following line in the opened file:
/usr/local/lib
sudo ldconfig sudo apt-get install libjpeg62
in order to calibrate the camera you have to use the Sampleviewer software provided with the code above. In Ubuntu 18 this will not work unless you install libtiff4, follow this steps to install it
If you use the kilogui the OHC will not work with the Smart Arena. Reboot the computer.
If the OHC is not working at all, then add the current user to dial out group and reboot the computer:
sh sudo usermod -a -G dialout <your-user> sudo reboot now