Drones have been used to map enclosed spaces before, including caves and mineshafts deemed unsafe for humans, by Australian researchers at the Commonwealth Scientific and Industrial Research Organization. They met with considerable success. However, this research used manually controlled drones and video instead of sonar. They are now investigating the use of sonar over video, but have yet to publish the results. I plan on using several sonar rangefinders to minimize the data my onboard computer (Raspberry Pi 3) will have to process. I had not previously considered cave mapping as an application, but it seems like a very reasonable goal for me to work towards, since flying indoors in rooms and halls is similar to flying in caves and mineshafts. Additionally, the research is at a stage in which my autonomous aircraft would provide the next step towards this sort of tunnel mapping. All existing aircraft of this kind are manual.
The Raspberry Pi 3 will be using the Maverick operating system, which has been used in onboard drone processing by past researchers. Notably, the creators of Maverick created a drone that would automatically land on a designated pad using a rangefinder and image processing. All commands were sent to the flight controller and drone using a program called Dronekit, which interprets python script as commands for the quadcopter. Sources that include the API’s and documentation for both Dronekit and Maverick can be found in the bibliography.
Current autonomous drones with limited accident avoidance are a burgeoning field. From package delivery to terrain mapping and research, autonomous quad and hex rotors have proved an incredibly adaptable tool. If this can be expanded to include indoor flight with a smaller aircraft, drones could be used for a host of new purposes. Autonomous indoor flight can be useful for reconnaissance, search and rescue, and mapping. A drone that can navigate its environment without crashing, or human involvement, would be incredibly useful, opening the door to further autonomous missions for drones. Accident avoidance like what I will be attempting to create is only the first step. It would allow for pilots to fly in enclosed spaces without crashing. However, an autonomous aircraft that generates its own map of the environment could be outfitted for a myriad of tasks, from mapping cave systems to searching for earthquake victims.