Tornyol builds 40-gram AI drone to hunt mosquitoes with sonar
Tornyol has built a 40-gram autonomous drone designed to detect and kill mosquitoes in flight.
The system was developed by Alex Toussaint, a former engineer at missile maker MBDA, through a Y Combinator-backed startup. It combines ultrasonic sonar, phased-array microphones and onboard AI to find small flying targets and intercept them before they enter a defined protection zone. The current prototype is priced at $1,100, though Tornyol says mass production could eventually bring the cost down to a few hundred dollars.
In an early test, the drone detected a 2 mm-wide plastic ball from 30 centimeters while using only 1/1300th of its planned transmitting power. Toussaint said the result validated the system’s signal-to-noise budget and moved the project closer to its target of tracking mosquitoes at three meters. The team also reported its first air-to-air kill after the autonomous drone intercepted a flying moth during an outdoor test flight.
Operation is built around a base station and a mapped protection area. A user defines the zone on the base-station map, starts the mission and does not need wiring or fixed installation. One Tornyol unit is designed to cover indoor or outdoor areas of up to 5 acres, but the current prototype is limited to three minutes of flight per charge before returning automatically to its dock for a recharge cycle of about 30 minutes.
The drone uses TDK InvenSense T3902 MEMS microphones in its sonar array and consumes less than 5 W, within USB 3.0 limits. An FPGA processes acoustic data from dozens of sensors in real time. Tornyol markets the device as a quiet, chemical-free alternative to pesticides, but the same combination of acoustic detection and autonomous aerial interception could have wider implications as demand rises for small systems that can track and stop drones.