Small UAVs (under 20 pounds) have several distinct advantages in the remote sensing field. First, they are cheap. An optimized, well-planned UAV that can carry out missions with flight times of an hour can cost as little as $5,000 but can go as high as $50,000 – although prices like these will certainly fall as competition increases. Compared to the cost of a full-scale plane or helicopter (roughly $200,000), the low end UAV comes at an extremely reasonable price, and the high end is still a quarter of the cost of a manned aircraft. Further, it costs around six cents each flight to charge a battery for UAV flights compared to the cost of fuel for full-scale platforms.
Secondly, cameras onboard UAVs can achieve a much higher resolution than conventional airplanes or helicopters. A UAV based camera can reach accuracy of one to three centimeters per pixel – compared to the meter per pixel accuracy of manned aircraft and satellite options. Finally, UAVs can be launched autonomously, meaning that an operator does not need any pilot background or training. From a tablet or smartphone,
anybody can successfully run a UAV remote sensing mission using open source software like APM MissionPlanner or commercial software that comes with some UAV systems.
The output of these missions depends on the sensors attached to the UAV platform – but much can be achieved with a common digital camera. With a generic point and shoot camera, action camera, or even an iPhone duct taped to the bottom of a UAV, we can create 3D models, angularly and geographically correct maps, and texture meshes. In agriculture, these outputs can be useful in our pursuit of maximizing profits.
3D models show the exact geographic aspects of the land. They can be used to find the height of a crop – so in the example of corn, a farmer could tell the height of his or her corn row by row, and identify an area in his or her field that is shorter than the average and act on this knowledge.
Angularly and geographically accurate maps are maps that can represent the exact geographic location of a point displayed on the map, and all distances on the map represent the relative distances on the ground. This is useful because with the image output, we can now know the exact latitude and longitude of our underperforming corn area and treat it directly.
Texture meshes are 3D models that have a mosaic of stitched images laid over them. They provide us with the most useful output, since they are full color orthorectified mosaic of the area being surveyed. Orthorectification means that the mosaic of stitched images is angularly and geographically accurate. So in agriculture, we can use this to identify discoloration of a section in our field (possibly due to pests) and know the exact latitude and longitude to investigate.
In order to produce these outputs, we need to use an aerial platform to take pictures of an area with overlapping sections, and then import the resulting images to several types of post-processing software.