A novel sound source localization method using a global-best guided cuckoo search algorithm for drone based search and rescue operations.

During disaster management, unmanned aerial vehicles play an important role as they are deployed for effective search and rescue operations. Nowadays, personnel from many emergency services like police officers, firefighters, and volunteers in rescue teams take advantage of drones for cost-effective and time-efficient search and rescue operations. All drone-based search operations to date either have used vision-based drones or require direct human assistance along with the drone. During low or almost no visible light conditions, as well as in foggy weather, it becomes nearly impossible for vision-based drones to continue the search operation. Thus, we propose an alternative approach for drone-based search and rescue operations using acoustic source localization. During the search operation, the sound source is the victim, who is screaming for help. The signal emitted by the source will be recorded using an array of microphones embedded in the drone structure. Thus a novel approach is presented to analyze the captured audio signals for accurate source localization. The proposed method deals with the ego noise and wind noise generated by the motion of drone, its motors, propellers, and other stationary structural noise. A global-best guided cuckoo search optimization algorithm is used to estimate the noise components present in an unknown noisy signal. After estimation of noise either temporal subtraction is used in proposed Algorithm 1 or the Wiener filter as in proposed Algorithm 2 to suppress the noise. Afterward, the time difference of arrival-based sound source localization (SSL) is carried out to compute the coordinates of the speech source. The results demonstrated that Algorithms 1 and 2 both can localize the sound sources effectively whereas conventional SSL algorithms failed to produce the correct results. Further, Algorithm 2 outperforms Algorithm 1 in terms of accurate position estimation and signal-to-noise ratio improvement.