SPECTRUM AGH TEAM win signal processing competition

The team of students from the SPECTRUM Student Research Team took first place in the prestigious international competition – Signal Processing Cup 2019.  The signal processing algorithms developed by AGH UST students proved to be the most accurate.

The competition was aimed at developing an effective method of detecting the location victims of various disasters by means of a quadrocopter an unmanned aerial vehicle (UAV) supplied with a set of microphones.

The competition was a three-stage challenge and required that the participants work out the direction of incoming sound wave emitted by sound sources using recordings from an 8-channel cubic microphone array positioned in the UAV. The teams started by locating static (speech) and then dynamic (white noise / broadband signal) sound sources. In Stage 1, both the source and the quadrocopter remained motionless during audio recording. Then, in the second phase, both were put in motion. Stage 3 involved location of speech as a dynamic source sound, with both the source and the quadrocopter moving while the recording took place. The tasks posed a considerable level of difficulty due to the high volume of acoustic noise produced by the UAV’s rotors positioned in very close proximity of the microphone array. This resulted in a very low power of the recorded sound source signal (barely audible for the human ear) compared to the power of the noise produced by the UAV. Acoustic noise generated by the quadrocopter consists of harmonic components associated with propeller velocity, stationary structural noise and the wind noise caused by the movement of the UAV and rotation of the propellers. An additional source of difficulties was the fact that the UAV was in constant motion, occasionally with sudden changes of direction, which led to more complicated trajectories of the sound sources that were to be effectively located by the system. 

The competition, which was held online and lasted from 11 November 2018 to 11 March 2019 attracted 20 teams from 13 countries. The SPECTRUM Team scored the total of 738 points, gaining the advantage of several points over the runner-up.

The winning team has been oficially supervised by mgr inż. Szymon Woźniak, PhD candidate at the Faculty of Computer Science, Electronics and Telecommunications. The other members included Mieszko Fraś and Mateusz Guzik  - acoustic engineering students from the Faculty of Mechanical Engineering and Robotics and Piotr Walas, an electronics graduate at the Faculty of Computer Science, Electronics and Telecommunications.

The AGH SPECTRUM Signal Processing Research Team is affiliated with the Department of Electronics at the Faculty of Computer Science, Electronics and Telecommunications. Jakub Gałka, PhD, BTech, acts as the team’s research supervisor.

The Signal Processing Cup (SP Cup) is held once a year and is organised by the  Institute of Electrical and Electronics Engineers, officially known as IEEE, which is an association of professionals dealing with electrical engineering and electronics as well as related domains. During the SP Cup, student teams attempt to solve real problems using methods and techniques of signal processing. The three winning teams take part in the world cup, where they compete for the main prize – USD 5000. This year’s cup will be held in Brighton, UK, during the international event IEEE ICASSP 2019.