BACKGROUND
The DC Air National Guard works closely with the Metropolitan Police Department and other government agencies to ensure large-scale events such as July 4th and the Presidential Inauguration remain safe to the public. The Guard provides critical support such as crowd management, traffic control, communications, logistics, media support, public affairs, and ceremonial duties. Large-scale events can produce an overwhelming amount of rogue drone activity, which in the event of an emergency, could lead to danger and injury to the public. For example, an overwhelming amount of rogue drone activity could cause unnecessary air traffic if emergency helicopters had to enter an area.
CHALLENGE
National Guard officers on duty during major events in DC need a way to disrupt the flight of a rogue drone in order to reduce the risk of danger during a large-scale event
LIMITATIONS
– Several targets and little regulation
Blog Posts
3/15/18
Over the course of the semester so far, we have been trying to narrow down the problem as well as find a viable solution. Based on our research, we have divided the problem of taking down a drone into four stages, which are listed and described in detail below.
- Detection: Detecting a drone is not a trivial task. Many current systems rely on radar and radio frequency(RF) sensors. Several companies also use other detectors such as cameras(DeDrone) as well as acoustic sensors. The main problem we have found so far is that these detection methods are largely ineffective in complex urban environments. Furthermore, they aren’t universally applicable to every drone.
- Identification: Detecting a drone is one thing, actually gathering useful information about it is a whole problem in itself. How do we know if a drone is carrying a biological weapon? An explosive? A camera? Our takedown method will probably differ based on what the drone is carrying. Companies like DeDrone are able to provide real time info about a drone’s heading, speed, model etc but not really about what kind of payload it is carrying.
- Decision: In between Identification of a drone and taking it down, a decision must be made. In our scenarios, we are assuming that we will in fact take the drone down every time, but how we do so may differ depending on the situation. Not many companies that we know of consider this step. We think this is because many companies are content with having one size fits all takedown method.
- Takedown: Actually taking down a drone can be done in a variety of ways. The main techniques are mechanical approaches (net, gun), RF interference(Drone Shield), and hacking(MESMER, Icarus). Each has their pros and cons, which we are looking into as we progress.
So where does this lead us? Now that we have a grasp on the process of taking down a drone, and the issues at each step, we are looking to develop, optimize, and integrate new and existing technologies to solve the task at hand. One of the most promising fields for this problem is the field of artificial intelligence. Neural networks combined with computer vision provide great detection and identification tools. Neural networks could also be used to build a Decision Support System that automatically makes the optimal takedown decision by considering various factors. As we progress farther into the semester, we hope to research how AI might be used to help defeat drone threats.
3/1/18
Another test blog post.