This setup presumes pre-installed ROS Kinetic with mavros and Gazebo 7 on Ubuntu 16.04
- Get the PX4 autopilot software
$ mkdir ~/src
$ cd ~/src/
$ git clone https://github.com/PX4/Firmware.git
$ cd Firmware/
$ git submodule update --init --recursive
$ make posix_sitl_default gazebo
- Make sure you have all the required gazebo models in
.gazebo/models/folder - Source the Gazebo Model path and PX4 flight stack
$ source ~/src/Firmware/Tools/setup_gazebo.bash ~/src/Firmware ~/src/Firmware/build/posix_sitl_default
$ export ROS_PACKAGE_PATH=$ROS_PACKAGE_PATH:~/src/Firmware
$ export ROS_PACKAGE_PATH=$ROS_PACKAGE_PATH:~/src/Firmware/Tools/sitl_gazebo
$ export GAZEBO_MODEL_PATH=$GAZEBO_MODEL_PATH:~/catkin_ws/src/warehouse_interiit/models
- Zbar ( for barcoe node )
$ sudo apt-get install ros-<distro>-zbar-ros
-
WiringPi ( For Sonar on RPi )
-
GeographicLib
$ sudo apt-get install libgeographic-dev
Once all dependencies are installed
- Clone and build this repo in your catkin workspace
NOTE : If the build failed to find the Line.h file , comment out the executables and target libraries from the CMakeLists.txt file and build the messegaes first then uncomment the executables and build again.
- Run the simulation
$ roslaunch warehouse_interiit px4_interiit.launch
$ roslaunch mavros px4.launch fcu_url:="udp://:14540@192.168.1.36:14557"- Once the Simulator boots up run the core nodes
$ roslaunch warehouse_interiit warehouse.launch
Config.hcontains all the constants and camera feed strings. Change the values according to the setup.- Setup the mavlink between the RPi ( or any onboard processor )
- Run the strategy, commander, line_detection and barcode node.
For Running OnBoard
$ roslaunch warehouse_interiit warehouse.launch
For Running OffBoard ,create a ROS_MASTER and create a multi PC ROS
$ roslaunch warehouse_interiit warehouseOff.launch #Onboard
$ rosrun warehouse_interiit line_detect #On PC
Published Topics :
/altitude_setpoint #Sends Target Altitude
/cgo3_camera/image_raw #Downward Facing Camera Video Feed
/code/bar #Barcode decoded Strings
/code/image #Detected Bar/QR Code image with bounding box
/code/qr #QR Coded decoded string
/code/scan #Boolean to signal the trigger of scanning
/feedback/horizontal #Current Target / Hoverig Line
/feedback/vertical #Current Vertical Following Line
/lines/horizontal #Array of Horizontal Detetcted Lines
/lines/vertical #Vertical Detetcted Line
/right_camera/image_raw #Barcode Scanning Side Facing Cam Feed
/state #String to tell controller to switch states
/edge #Boolean to publish vertical edge line array in alternate algorithm
/square/vertical #Publishes vertical line from centre of the square landing area
/square/horizontal #Publishes horizontal line from centre of the square landing area
Line Stores rho and theta of the lines.
Linearray Stores the array of Lines.
Color segmentation based on HSV color space is done on the down ward facing camera to locate the lines for quad to follow. Whole image is divided into N_SLICE_H x N_SLICE_W ( declared in Config.h ) parts and contour detection is applied on each one of the segment to form multiple circles outlining the grid lines. Hough Line is used on the circles to find the lines passing through the circles in a binary image.
Clustering Algorithm is used to find the best fitting line and publishes the line in vertical and array of lines in horizontal direction.