In the last decade, UAVs, or unmanned aerial vehicles, have blown up in popularity. No longer an advanced tool for the military, today’s drones span an incredibly wide range of uses on the individual and commercial level. The most common uses for UAVs are in surveillance and cinematography, due to the fact that deploying small planes without human cargo is both cheaper and safer. As the prices of small drones drop and their power and battery life increase, more and more specialized applications will arise. Ecology in particular is a field in which time-intensive ground surveys are commonplace. Similar to the Harvey Mudd Bee Lab’s own drone-mapping project, many other research organizations are using UAVs to capture aerial images quickly and cheaply to later be processed and analyzed.
At the University of Sydney, researchers are using drones to create large-scale maps of the ground and classify different species of trees and vegetation. Their drones are equipped with a GPS unit as well as an IMU (inertial measurement unit) in order to estimate the location of the drone in reference to the global frame of the Earth. After programming the drone to fly in a “lawn-mower” pattern (see Figure 1 below), the images were stitched together into 3D terrain maps.
| Figure 1: Lawnmower flight path. |
The research team used multiple image processing techniques to identify tree and vegetation species including SVM (support vector machines), RGB intensity, and texture analysis. The results of their classification process are displayed in the figure below.
 |
| Figure 2: Tree species classification from the University of Sydney research team [1]. |
Researchers from the University of Tasmania and the University of Wollongong have travelled to Antarctica to analyze moss beds to study the effects of climate change. Past methods for analyzing ground data include using satellite images or ground images (taken by hand). This research team is using small UAVs to quickly and cheaply create large maps of moss beds at high resolution. The team built maps and then used elevation data to characterize the expected conditions of the moss in different areas (which are affected by solar irradiation, slope, and more). By comparing the expected moss conditions with the actual images from the drone, the researchers drew conclusions on how climate change played a role in the growth of moss. Figure 3 displays one of the maps that were stitched together using the images taken from the quadcopter.
 |
| Figure 3: Map of Antarctic moss beds [2]. |
Another key application for UAVs in conservation is surveillance, which is the focus of two organizations in South Africa and Kenya. These groups are not working for a military or trying to capture terrorists; instead, they are trying to save the diminishing population of elephants and rhinos.
A team at the University of Maryland has built an anti-poaching system based out of South Africa to reduce the slaughtering of rhinos and elephants. They create probability maps for the possible locations of animals using satellite data and historical data, and deploy rangers and UAVs to locations where they expect to find poachers. The UAVs patrol the area quickly and efficiently, and using an onboard camera, information is streamed back to the rangers. The rangers can then move to the exact location of the incident and arrest the poachers on the spot. The UAVs provide the crucial role of surveying a large area of land quickly, which is important since the team cannot provide an exact location for a probable poaching incident from their algorithms.
 |
| Figure 4: Anti-poaching drone flying in South Africa [3]. |
In Kenya, the Ol Pejeta Conservancy has received help from Airware, a drone autopilot company out of San Fran, to combat poaching with UAVs. Their system is still in development, but they have gone through a successful testing phase and are working on deploying their drones in the near future. Ol Pejeta is using larger fixed-wing UAVs with thermal imaging cameras. Their hope is that they will spot poachers on the thermal camera stream at night where a normal RGB camera lacks useful information. They are also working on software that will use image processing to automatically identify animals on the images streamed from the drone.
We have only begun to witness the vast range of applications for UAVs. In the next decade we will likely see drones everywhere, performing tasks such as delivering packages and seeding fields. The introduction of drones to biology, however, is an even more exciting prospect. Ecological fieldwork has historically been an extraordinarily tedious procedure, involving lots of man-hours. Replacing human feet with propellers and motors will allow researchers to obtain more data, and the latest advances in image processing and machine learning allow research teams to analyze that data quickly. The projects described above give just a taste of how UAVs can address problems in conservation and climate change: what’s more exciting is the future applications in ecology as drone technology continues to improve.
Sources
Bryson, Mitch et al. (2014) “Cost-Effective Mapping Using Unmanned Aerial Vehicles in Ecology Monitoring Applications.” Experimental Robotics. Ed. Oussama Khatib, Vijay Kumar, and Gaurav Sukhatme. Springer Berlin Heidelberg. 509–523.http://link.springer.com/ chapter/10.1007/978-3-642-28572-1_35
Lucieer, A, Robinson, S & Turner, D (2010) Using an unmanned aerial vehicle (UAV) for ultra-high resolution mapping of Antarctic moss beds. 15th Australasian Remote Sensing & Photogrammetry Conference. Alice Springs. 14-16th Sept 2010.
http://ro.uow.edu.au/cgi/viewcontent.cgi?article=1476&context=scipapers
South African anti-poaching team.
http://www.slate.com/blogs/wild_things/2015/01/28/drones_for_wildlife_conservation_rangers_uavs_and_math_protect_elephants.html
Kenyan anti-poaching team.
http://www.bbc.com/news/business-28132521
Further Reading
Platforms for aerial mapping.
http://vterrain.org/Imagery/self.html
Applications of drone-mapping using the Micro Drones platform.
http://www.microdrones.com/en/applications/areas-of-application/science-and-research/
University of Washington’s introduction to texture analysis.
https://courses.cs.washington.edu/courses/cse576/book/ch7.pdf
Media Credits
[1] Figure 6 from Bryson et al 2014.
[2] Figure 6 from Lucieer et al 2010.
[3] Photo by Thomas Snitch.
http://www.slate.com/content/dam/slate/blogs/wild_things/2015/01/28/drones_for_wildlife_conservation_rangers_uavs_and_math_protect_elephants/image201501262453117okr76.png.CROP.promovar-mediumlarge.png
No comments:
Post a Comment