Authors: Gabe Snow, Joseph Ward
Mentors: Cammy Peterson
Insitution: Brigham Young University
Unmanned aerial system (UAS) research is fast becoming an important area of development in commercial and military applications. As UAS become more prevalent commercially and recreationally, there is a growing need to accurately track large numbers of these aircraft. This is particularly important in compact urban environments where potential flight paths are limited.
Our research team at BYU is developing the Local Air Traffic Information System (LATIS), allowing multiple radar ground stations to communicate and track UAS across multiple fields of view. One major component of this system is Recursive Random Sample Consensus (R-RANSAC)---an algorithm used to correlate and combine the data from multiple sources.
The process a ground station uses for calibration is to collect Real-Time Kinematic Global Positioning System (RTK GPS) coordinates of a friendly UAS as it passes through the field of view of a station's radar. R-RANSAC is then used to a) filter noise from the raw radar data, and b) identify "tracks," or paths which UAS have followed, using temporal and spatial proximity. The Orthogonal Procrustes Problem then provides a method to rotate data from the local radar frame to the global frame. These steps can be done live or with recorded data. Following this, the calibrated radar uses R-RANSAC to filter data and identify passing UAS with high accuracy.
Our contributions to the project are developing communication software, refining R-RANSAC, and helping to implement the whole system in flight experiments. We are continuing to work on analyzing the data taken from flight tests and publish the improvements of this system compared to existing methods.