Drone Search and Track

Imagine a drone with thermal and video cameras in a sanctuary. The drone can be programmed how it moves, but also the cameras can be panned and tilted to extend the view. How can such an aerial surveillance system be used to search and track on the savannah?

Tracking

Consider a scenario where the goal is to keep track of both rhinos and vehicles travelling on the roads inside the sanctuary. In the scenario below, there are three rhinos standing still and three moving vehicles. The search and track of the drone is programmed to maximise the information about these six vehicles at the same time as minimising the risk that an object is lost.  The animation shows the six objects, the road network, how the drone is moved over the terrain, and how the sensors are moved. The footprint of the current camera view is marked with green, and the most recent footprints are shown in red. The pan and tilt of the cameras are rather fast, which relaxes the requirement of the drone motion. 

Animation by Per Skoglar as part of the ARCUS project, and documented in
Road Target Search and Tracking with Gimballed Vision Sensor on an Unmanned Aerial Vehicle
P. Skoglar, U. Orguner, D. Törnqvist and F. Gustafsson
Remote Sens. 2012, 4(7), 2076-2111 

Searching

Before the drone is put in a tracking mode in a completely unknown environment, it has to explore the area first. The drone and camera are steered to optimise information in a certain time horizon, which makes the trajectory unpredictable and systematic at the same time. The seemingly random motion makes it more difficult to hide for intruders, compared to the more natural systematic zig-zag motion.

Animation generated by Per Skoglar in the ARCUS project, and documented in
Information based planning for aerial exploration. Technical Report LiTH-ISY-R-3047, Department of Electrical Engineering,
Linköping University, 2012