Contact: Dalia El Badawy, Dümbgen, Frederike
Synopsis: Study and test drone’s capability to localize itself from acoustic signals.
Level: MS
Description:
Although it is often seen as a nuisance, scattering of sound — acoustic waves “bouncing” off physical objects — has been successfully leveraged to determine the direction of arrival of sound sources using only one microphone [1]. Motivated by this success, we would like to explore the usability of sound signals for self-localization of drones inside cluttered rooms.
This project consists of 3 parts:
- Prototyping of a sound recording pipeline for multiple microphones mounted on the crazyflie drone (https://www.bitcraze.io/crazyflie-2-1/)
- Recording and processing of the drone’s audio response at a set of points inside a room
- Algorithm design for localization from audio, using a physics- or learning-based approach based on the results from point 2.
[1] D. El Badawy and I. Dokmanić, “Direction of Arrival With One Microphone, a Few LEGOs, and Non-Negative Matrix Factorization,” in IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 26, no. 12, pp. 2436-2446, Dec. 2018.
Deliverables: A report and a working system with clear documentation.
Prerequisites: Interest in audio signal processing, solid programming skills.
Type of Work: 30% hardware programming in C, 40% algorithm design in python, 30% experiments and evaluation