Listener Drone

The Listener Drone (both RC operated or Autonomous) is a unique drone system with the ability to listen to sounds, the way humans do. Similar to how we can locate the origin (i.e. direction angles, distance) of a sound just by listening to it, Listener Drone can also locate the sound source in real-time with high-precision. This hearing capability of the drone is helpful to complete tasks, where auditory cues are the only possible resources. For example, Search and Rescue operations in low-visibility conditions. The Listener Drone can also be used for Animal Call based Census, From the sky Anti-poaching, Rescuing hikers, travelers in forest (as seeing through the canopy is difficult), help in audio-based drone navigation where GPS unavailable. The entire system is faster (>= 2 times) and cheaper (1/8th the cost) compared to traditional vision-enables drones.

​

Applications: In low-visibility conditions like dense fog, smoke, wildfire, heavy rain, inside coal mines; during night time; in occlusions after earthquakes, avalanches, landslides; through forest canopy. 

​

Timeline: January, 2019 - June, 2021.

​

Collaborator(s): WeRobotics (USA)

​

Awards & Honors: 

​

• Received Research Grant INR 6,00,000 from Department of Telecommunications, Govt. of India for integration of 5G technology in our drone system .

• Received Research Grant USD 15,000 from WeRobotics for prototype development and testing.

• Selected among the top 9 Global Finalists of Unusual Solutions Competition. 

• Invited exhibitor at the India Mobile Congress 2020, South Asia's largest Digital Technology Event. Invited by the Department of Telecommunication (DoT), Government of India. 

• Invited Presentation at Final Pitch Event, organized by WeRobotics at Nairobi, Kenya.

• Secured Global Rank 13 in IEEE Signal Processing Cup 2019. 

 

Media Coverage: 

• By ABP Education

• By Bartaman Patrika

• By Ananda Bazar Patrika

• By Aj Kal Patrika

 

Publication:

Banerjee, A., Nilhani, A., Dhabal, S., Venkateswaran, P., 2021. A novel sound source localization method using a global-best guided cuckoo search algorithm for drone-based search and rescue operation. In Unmanned Aerial Systems: Theoretical Foundation and Application, pp. 377-417, Academic Press. [PDF]

 

Project Website: HERE

 

Theory behind Working: The drone embedded hardware continuously captures the surrounding sounds, which is then sent to the processing unit. The processing unit has mainly two tasks: 1) Dealing with the ego noise produced by drone motors, propellers and the wind. 2) Extracting directional cue to locate the speech coming from ground direction. The most critical part in the process is the Ego noise. Being very high in amplitude it makes the Signal to Noise Ratio (SNR) as low as -15 dB. And it overlaps with the speech frequency spectrum, thus application of classical adaptive filtering is inappropriate. In our work we combined an optimization method along with the angular spectrum based (Generalized Cross Correlation) Sound Source Localization to locate speech sources precisely. Also the hardware implementation is such that this drone system is at least twice as fast as the traditional ones in locating victims, who is shouting for help. 

​​