Guardian Angel Demo
Guardian Angel's main goal was to prove the ability of autonomous navigation of a quad-copter indoors.
As you can see from the following video demonstration we imagine ourselves a use case where a small fleet of quad-copters would greet every visitor right at the entrance to the building, where each visitor would have the option of requesting an escort to an unfamiliar part of the building. Imagine finding a relative in a huge hospital complex, or shop in a mall, or your professor's office on the university campus.
In this specific system, the quadcopter is “blind” and solely relies on external cameras and a server to guide it.
The guiding system for the quadcopter is composed of a network of external cameras deployed along the possible routes, each camera is connected to a small IoT device with minimal computing power, which in turn via a network is connected to a server that runs most of the computations.
In addition to the quadcopter guiding system, we employ another system that is tracking the human visitor. This way, we can keep the quadcopter at a reasonably safe distance from the visitor and not lose him at the same time.
The system is capable of storing multiple routes and users can ask how to get to any mapped part of the building.
It is important to note that most of the quadcopters today are not capable of indoor navigation, and this is because they rely on GPS for navigation, which is not available indoors. Those drones that do offer indoor navigation due to the size of the required sensors usually are not legal for indoor flight.
So there you have it, a cutting-edge proof of concept project combining big data, core sets, computer vision, networking, and a big chunk of math.
Contributors:
Soliman Nasser and Ibrahim Jubran - founding fathers of the underlying algorithm and system
Artem Barger - Apache Thrift voodoo master
Mike Vologin - Python ninja
Karina Turevsky - GUI expert
George Kesaev - project owner, designer, developer, and the glue that got this thing working.
Comments