Drone_TLD (shadow) : A tracking and following application for UAVs using OpenTLD

Packages Required

• ardrone_autonomy

Installation

• Download shadow package and Drone_TLD. Drone_TLD is the rosified verison of the OpenTLD by Zdenek Kalal.
  git clone https://github.com/HALLAB-Halifax/Drone_TLD.git
• catkin_make the shadow package.
• You need to add ROS support to MATLAB for Drone_TLD to work. Refer to ROS support for MATLAB to install it.

Get it running

• Connect to AR.Drone's network and initiate roscore.
• Run ardrone_autonomy to setup the topics to communicate with the AR.Done.
  rosrun ardrone_autonomy ardrone_driver
• Run the file run_TLD.m of Drone_TLD on Matlab.
• In the window showing the camera frame draw a bounding box over the target.
• Next run shadow to get the drone to follow the target.
  rosrun shadow shadow
• Watch the drone follow the target. Have Fun! :)

Troubleshooting

Drone_TLD

• Compile.m: Make sure the veriables include and libpath point to the right directories. If unsure, use the terminal to get the paths.
  pkg-config --cflags opencv for include
  pkg-config --libs opencv for libpath

• tldExample.m

  1. Specify the topic to subscribe from for the camera feed.
  2. Two callback functions are available. Use functionrgb when the subscribed topic is color else use funcBW for grayscale input.

• For further info or toubleshooting on TLD refer to the issues section of OpenTLD