It is difficult to dismiss how much drones have disrupted society in the last five years. They map and monitor the land. Hollywood shoots movies from the platforms. Power companies inspect high voltage lines with the machines. As a result of this widespread industrialization, drones have become so cheap that you might even have one in your garage.
It is also easy to overlook that unmanned aerial vehicles are not a new invention. The first ones were literally remote controlled World War 2 aircraft modified to deliver bombs to targets. After the war, the derelict airplanes found use as missile targets. Later during the Vietnam War, specialized drones ran reconnaissance missions over hostile territory. In the years since, the military has added decoy missions to the list of drone jobs.
Outside the military, the catalog of applications also keeps increasing. You might not even know that a drone (UAV) could be above your head right now, spying on you. Before you run screaming from whatever building you are in, don’t do that if you are near a construction site. Every year about 150,000 injuries happen on construction sites, and about a thousand incidents are fatal.
One company striving to change this is DroneDeploy. They don’t own or operate any aerial vehicles. Instead, the firm analyzes and stores images uploaded from a drone. Their typical customer is in the construction, mining, or agriculture industry. Users provide a Dà-Jiāng (DJI) manufactured aerial vehicle and a mobile device. DroneDeploy furnishes the application for uploading, retrieving, and analyzing the map data.
With (DJI) aerial vehicle prices running at about five hundred dollars, the entry to service is not trivial. DroneDeploy subscriptions begin at about one hundred dollars a month and include a range of mapping and analysis features. Put in perspective, it can cost in the range of three hundred dollars an hour to hire a cheap helicopter for photography and mapping jobs. Tasking a satellite for high-resolution imagery can be astronomical at nine hundred dollars to start. Archived photos run at a quarter of the cost but are also generally useless for monitoring ongoing projects. Satellite imaging only makes sense when your only option is mapping an entire county or province.
Returning to construction, the technology has already been successfully demonstrated its usefulness at a site in Cushing Oklahoma. The city is a hotspot of crude oil storage, one of the largest in the nation. When it became necessary to build additional storage tanks, managers needed regular photography to measure the progress of fabrication. Just as in manufacturing, quality assurance is important. When the dykes were excavated around the vats, precise clearances are needed to contain spills. Typically surveyors measure the dimensions, but now the drones can verify their work.
Drones are also getting into commercial aviation, and not just by flying in the path of the jet you are aboard right now. Before you panic and try to force the door open, Intel and Airbus are teaming up to inspect newly built aircraft.
In the primitive past, engineers and technicians had to physically inspect aircraft at the end of production. They probably also had to inspect the jets while on the manufacturing line too, exhausting.
At the Farnborough Air Show in England, a team of engineers from Airbus recently demonstrated visual aircraft inspection using an AscTec Falcon 8 drone. The Intel New Technology Group provided the RealSense camera.
The Falcon 8 drone was operated by a human pilot and followed a predetermined path over the jet. The camera was set to photograph automatically and captured a mosaic of images that were digitally reconstructed into a 3-dimensional model. In the manufacture of new jets, Airbus is primarily interested in any scratches, dents, and painting defects that might diminish a completed aircraft. Serious mistakes are usually caught long before the final inspection. Any documented defects will be forwarded to a database for improved traceability and monitoring.
The Asctec Falcon 8 drone is a patented V-form octocopter fitted with a 42-megapixel camera and a RealSense sensor. The RealSense device tracks obstacles while the camera photographs the jet.
Airbus plans to examine the jet at the end of production, in a process that should reduce inspection time from two hours to fifteen minutes. The company is presently using the Airbus 330 aircraft as the pilot vehicle, but there are plans to fully develop the technology for all of Airbus’s aircraft.
The Intel RealSense camera is actually a composite of four sensors, using a conventional camera, infrared laser projector, an infrared camera, and a microphone array. The RealSense camera features depth perception, that it achieves by observing an infrared grid projected from the laser. Outside of 3-dimensional modeling, the sensors can also be used for facial recognition and gesture recognition. The camera features a resolution of 1080p and can have a resolution as high as 0.2 meters.
The camera achieves this feat by analyzing incoming information in real-time with a Vision D4 Processor. This is itself a specialized graphic processor. The processor uses preset algorithms to transform the raw images into 3-dimensional maps, offloading the task from the platform that it is attached to. The RealSense camera itself is sold by Intel as a module that can be plugged into another computer, robot, or VR system with very little development. The modules can also be customized for power consumption, field of view, and shutter type to optimize the needs of the application.
Outside of drones, the cameras have applications in Virtual Reality as eyes for the user. The instruments have also been integrated into laptop computers for the gesture control of video games.
The cameras are also being investigated for medical uses, as a new eye for the blind and disabled. Users might also appreciate the instruments as an assistant to video chatting, where the computer’s ability to detect the speaker from background noise is important.
In the end, the accepted robotic contribution to manufacturing with dumb cumbersome machines is very outdated. The new generations of machines are small, quick, and smart. Most importantly, the new robots won’t take your job. They will make it simpler.