Ultrasonic Localization of a Quadrotor Using a Portable Beacon

Abstract

This thesis presents a method for localization of a quadrotor using ultrasound with five receiving nodes and a portable beacon. The time of flight is measured from when the ultrasonic signal is produced to when it is received and triggered through threshold detection. Two different lateration algorithms are explored to determine the position, the analytical trilateration and linear least squares method. The linear least squares algorithm outperforms the analytical trilateration method for static position testing and GPS-denied hover. A new approach named the combined lateration algorithm combines these two methods and has the best stationary standard deviation with 8.0 cm within a one metre radius at a height of 1.77 m. It calculates a successful measurement 93% of the time, which provides an average update rate of 9.3 Hz.Successful autonomous hover and landing is performed using a stationary and moving beacon within a GPS-denied environment.iii There are so many people who have helped me through my journey at Carleton University. To start, I would like to thank Carleton University for providing a fantastic learning experience through my Undergraduate and masters degree.A massive thanks goes to Professor Jason Etele who has been an excellent mentor throughout my degree. I'm still amazed at the ability to break down a problem to its roots and solve it methodically. I always enjoy the presentation dry runs where you completely reorganize presentations in a clearer manner. I am thankful for the research opportunity you provided through a project I am genuinely interested in with the ability to grow and learn new skills. The opportunity you provided to travel to San Diego and present at AIAA was phenomenal. The rowing chats are fun and have given me a benchmark to strive for when I take up rowing.A huge thanks goes to Ali Raza for the continuous support and help throughout my degree. From the very beginning you were excited to explain everything about TARA and your patience was incredible. You were always ready and willing to video chat when I was running into problems and provided great insight. Your experience with controllers was incredibly helpful while tuning my own PID controllers. You provided me with a great start with access to TARA 1 and a preliminary version of the ultrasonic localization system. Many thanks to Mark Sutherland who designed and built TARA 3 which I used for most of my testing. Thanks for providing me ideas on the first steps on integrating iv the ultrasonic localization system on TARA 3.A special thanks goes to the RUAS group, more specifically Brian Lemky and Craig Osborn. They developed the idea of the ultrasonic localization and provided a first version of the system.A massive thanks to Nagui Mikhail, who without I'd still be playing around on a breadboard. You were a massive help with developing my first PCB and troubleshooting the issues along the way. Beforehand I didn't even know how to solder or even what a PCB stood for.

RAIL: Robust Acoustic Indoor Localization for Drones

“…To report an overall average localization error for RAIL in any possible scenario and compare it with the state-of-theart, we conducted ample simulation and experimental tests with different trajectories with random paths and in various environments. Based on our evaluations and the overall report of the other work in the literature [2], [8], [14]- [16], RAIL achieves significant improvement in comparison with the previous drone localization schemes. For instance, in [2], their approach incurs a high Z-axis estimation error, and they did not propose any solution to fix it.…”

“…On the other hand, [15] has a localization error of at least 2 cm only for two dimensions without even addressing three-dimensional localization. In [14], the proposed scheme has an average error of 5.2 cm for three-dimensional localization for drones which is more than three times what RAIL achieves.…”

“…Thus, we decided to place the receivers on the drone while keeping the transmitters in the room. Another observation is that having one transmitter in the room and multiple receivers on-board the drone raises several issues [14]. First, it adds extra weight to the drone and increases power consumption.…”

PILOT: High-Precision Indoor Localization for Autonomous Drones

“…PILOT achieves significant improvement in comparison with other drone localization schemes in the literature [17], [18], [20], [42], [55], [56]. For instance, in [42], Famili et al failed to solve the Z-axis estimation error and their scheme had a much greater Z-axis estimation error compared to the X − Y plane estimation error.…”

“…In [20], Mao et al proposed an FMCW method to overcome the impact of interference and multi-path; however, their system was designed to track the drone on one line, just a single dimension, whereas PILOT proposes a three-dimensional localization. As another example, in [55], O'Keefe et al proposed a scheme for three-dimensional localization of drones which incurs an average error of 5.2 cm which is approximately five times worse than what PILOT provides.…”