Abstract:
Fruit ies pose a signi cant threat to fruit industry. Fruit infestation has e ected the
quality and quantity of our fruit export, reasulting in fruit ban and economical losses.
Traditional methods used to guard against these enemies seem ine cient. Techniques like
removing rotten fruits from the farm elds, using predators like ants and male sterilization
are manual and resource demanding methods. Luring the ies using traps with honey,
pleasant smell or di erent lights are passive techniques.
In this work, we proposed the development of a lightweight acoustic sensory mechanism
to be used on nano, low cost, drones against these fruit ies. We have demonstrated
that audio data, captured from simple microphones, can be used to discriminate between
helpful and harmful ies using machine learning. Additionally, the same audio data can
be used to localize the ies. The two above mentioned methods can enable us to track
the harmful ies.
We have used our acoustic sensory mechanism in indoor and outdoor environments
for the day and night experiments. Our nding of these experiments conclude that, if
training and testing both are done in real time environment, classi cation accuracy will
get better. For the drone experiments, we concluded that, due to low Signal to noise
ratio, signal processing techniques are not enough for drone noise removal. We need some
physical setup with light weight hurdles between audio sensing device and drone. With
the use of these physical hurdles, we have improved our classi cation accuracy but this
is not a very optimal setup for a drone to y with. We need noise damping drones, if
we want to proceed with the idea of nano drones in future. For now, these audio sensing
devices can be used in a form of an array to detect, classify and localize the fruit ies.
