The DEL project aims to create a system that takes real-time inputs from a flying drone and output an accurate state estimation with a 3-D error ellipsoid visualization. The system aims to achieve the goal of growing global commercial drone delivery operations safely and efficiently. Flying in dense urban airspace has risks that can include physical damage to humans or private property. An increase in the confidence of the position of a drone can help mitigate risks. Drone operators can see and react to collisions or other safety risks in real time, with increased accuracy. Supplying drone operators with accurate 3-D visualized data (error ellipsoids generated from location, speed, and heading data) will improve safety within urban drone airspace. Additionally, 3-D visualization can be utilized between multiple drones. If two autonomous drones are able to communicate with each other, they can identify the error ellipsoid around each other and better avoid one another. This capability will allow for more drones to fly closer together with higher confidence that no collisions or property damage will occur.
Drone speed and position are currently determined through a variety of sensors on the drone (GPS, IMUs, etc.). These sensors are accurate, but they are not always reliable (GPS denied environments) and each one has a different level of accuracy. Our approach is to apply an extended Kalman filter (EKF) algorithm to take real-time inputs from the sensors onboard the drones and output a 3-D state with 3-D error ellipsoid visualization using sensor fusion. The EKF creates an initial position vector from the given inputs and iterates. The iteration step updates the position and outputs an error estimate which will be used to create the error ellipsoid. Finally, the EKF outputs the best estimate to the PID controller to stabilize the system.
Kalman filter techniques are currently being used for determining the state of orbiting spacecraft, but they are also beginning to be applied to aircraft and drones. There is a lack of studies regarding using EKF on multiple drones while 3D modeling the error ellipsoids for collision avoidance. Implementing an EKF on multiple drones will help with tracking highly maneuverable drones in a dense drone environment such as a drone port. The drone delivery industry has the potential to grow to a multi-million dollar industry in the next 5-10 years. This technology could save companies millions of dollars when it comes to replacing damaged drones, equipment liability, or even lawsuits. State estimation data can later be used to create a centralized Air Traffic Controller for drones that could be almost entirely autonomous.
