Drones have been the center of attention recently in the news, especially from their usage as a military tool. The United States war efforts have credited much of their recent success to the use of large military drones to track and target hotspots.
Real questions began to surface recently when use on American territory could possibly track, target – and kill a potential terrorist without any approval process or consideration of citizens rights. It became a position of shoot first and ask questions later. The military seemed destined to use the drone as a tool and become judge, jury and executioner. Luckily the State department clarified its position and the use of drones on American territory. The issue seems closed for now. But the use of drones is now so important for the military that they currently train more drone-flying pilots than regular pilots.
Drones as theyre commonly called or UAVs (Unmanned Aerial Vehicles) have shown applications in other areas as well. They are being used as a tool for border control, fire/safety/rescue missions, aerial inspections of power lines, and yes even on large scale systems for aerial photography and LiDAR scanning.
I first became interested in drones, when I saw the Parrot AR-Drone at an electronics conference. This was a hobby device that looked like a Quad-copter about 24 inches square. It had two cameras one was forward facing and the second was facing down. It was controlled by the use of an iPad, iPhone, or iPod touch (iOS devices were originally used, but they now work with Android). The iOS device would connect to the quadcopters small computer thru a point-to-point Wi-Fi connection and the video could be streamed live to the iOS device. Or it could be captured for playback later.
You controlled the movement of the drone by tilting the iOS device, or used gesture controls and buttons on the iPhone. The company released an API so you could develop other applications to control this drone and the market really began to takeoff. The smooth control and movement of the device was amazing also. It was agile and quick, and could even flip smoothly in the air, follow a straight path and land softly. You did need to learn how to pilot the device to make it run smoothly.
Back in early 2010, when I first ordered this for the lab where I was working, there were some real doubts about the applicability to the construction space. It seemed like a fun toy, but how could it be used for applications. I started to show it at various conferences- FIATECH, Construction Industry Institute, Bentley Systems Conferences and the more I showed it, the more comments and ideas for applications began to follow.
Some of the ideas were simple but effective. Aerial inspections of areas that are hard or dangerous to reach. Photo/video capture of daily site project areas. Could the video clips be captured and used for photogrammetry? Could a LiDAR device be added to the device? How far can it fly? What was the image resolution? Can I add GPS? Can I program it to capture a large area by following sweeping patterns that I define? And finally what was the cost? The AR Drone is about $300, but could it be improved slightly with a modest increase in price and become a tool for quick and dirty LiDAR/Image scanning? Was this destined to become a common tool for LiDAR and image capture that could improve and automate the data collection process?
Skip forward to late 2012, and the reality of drones being used for LiDAR and image capture is now starting to come out of the universities and research groups to be applied in the field. The ability is beyond what I could imagine would be available today.
First that toy company has created a market in itself with many similar knock-offs trying to mimic its success by producing smaller, cheaper models. Second and more importantly, the French company, Parrot, the parent company of the AR Drone invested $7.5 million dollars in 2 software companies senseFly and Pix4D.
From their press release the senseFly drones are described as:
senseFly drones demonstrate the full extent of their capabilities when combined with software developed by Pix4D, a spin-off of EPFL’s Computer Vision Laboratory (CVLAB). The algorithms developed by Pix4D make it possible to assemble hundreds of aerial pictures and, within minutes, to get an extremely accurate 3D and geo-referenced image of the observed terrain
The company is combining these drones with software startups and doing some of the things that we only imagined when working with the original device. In fact, the press release also talked about many of the applications we originally imagined:
Mapping a construction site, landscape, cityscape or mine, monitoring industrial sites, managing agricultural land… The drone developed by senseFly, the result of 6 years of research in the Laboratory of Intelligent Systems at EPFL, facilitates the work of experts across various fields.
The toy is quickly becoming a professional spatial data capture platform. As the research and investment continues not only are witnessing the consumerization of drones we are also seeing these small scale devices being used for 3D data capture. This is a fast moving field. I believe we are on a similar path to creating low cost scanning tools for industrial quality work.
But this is just the beginning. For example, Dr. Vijay Kumar from the University of Pennsylvania and the GRASP laboratory is working on Autonomous Agile Aerial Robots. His work on handheld UAV robotics shrinks the drone size down, and uses a collection of multiple drones that are synchronized to capture a series of multiple images at the same time. The synchronized work of multiple drones working together in an area is known as swarming. His work has demonstrated indoor mapping is possible using this swarming technique. See the TED video link for details on his work and the application to indoor scanning. (I must admit the musical demo of the James Bond theme is interesting also.)
Another application where UAVs are being put to work is mining. The 3DM Analyst Mine Mapping Suite from Adam Technology can be used with UAV collected imagery. The DM Mining Analyst Software has a workflow defined with four steps: capture Images, determine camera orientations, generate DTMs and analyze data.
Images are captured in strips by multiple passes over the area with the UAV. This works for mapping terrain, stockpiles and open pit areas. For this to work you need a clear view of the entire surface, near uniform point density and accuracy, and high acquisition speed. In order to determine camera orientations 3DM uses CalibCam software.
To generate 3D data we need to determine where the camera was when each image was captured. Trying to determine camera location and orientation directly is difficult and even costly. Its easier, more accurate and robust to establish camera orientation by inspecting the images themselves. 3DM CalibCam can determine the relative orientations automatically, but cant get absolute coordinates. To establish a co-ordinate system (absolute orientations) it uses control points.
Finally, the 3DM analyst software can then generate the DTM model. This video clip shows the steps more clearly.
Hobbyists and researchers continue to work in this area to further reduce the size of the UAVs, integrate cheaper and better sensors, develop better software to automate the data collection, processing and management, as well as improve the battery and flight time for the drones.
This is creating a new class of nano-drone products like the DragonFly that are as small as an insect. Advanced Scientific Concepts has worked on developing small, light weight LiDAR scanners. These small LiDAR components are available now from Advanced Scientific Concepts with the TigerEye 3D flash LiDAR camera.
There is even a $99 Kickstarter crowd sourced project started last year to work on making a low-cost devices for general UAVs to use for LiDAR scanning.
Some of the more interesting challenges in this area still need to be resolved. The FAA needs to consider current rules and regulations Battery life and flight times need to be improved. The tradeoffs between using a swarming approach versus a single device need to be explored. In any case, drones are no longer pie-in-the-sky toys, and you can probably expect to see them overhead at sites you might be working on in the not too distant future.