Habituation Prevention in the Use of Ultrasonic Fruit Fly Control Systems using the Scare-Path and Shepherding Algorithm

Abstract

The cultivation of mangoes has emerged as a significant cash crop in Ghana, driven by both domestic and international demand. However, farmers face a significant challenge in the form of fruit fly infestations. Current mitigation strategies include the use of rubber bags to protect flowering plants and the application of pesticides. The former can sometimes reduce yield if the bag is not removed correctly or at the appropriate time to facilitate pollination. The latter is increasingly discouraged due to environmental concerns. In response to these challenges, we propose an ultrasonic control method for fruit flies. However, continuous ultrasonic broadcasting can lead to habituation, rendering the method ineffective over time. To overcome this, we introduce the "listen before broadcast" technique, supported by our novel "scare-path and shepherding" algorithm. This approach involves a sensing node that detects incoming fruit flies based on the frequency of their wing flaps and other physical or behavioural characteristics. The sensing method is inspired by the high-clutter prey identification strategy employed by foraging bats. The introduction of our habituation mitigation method, combined with the scare-path and shepherding algorithm, has yielded promising results. According to our simulation data, less than 10% of fruit flies were able to penetrate the inner perimeter of a mango plantation. These findings represent a significant milestone in the ongoing research into habituation prevention, with potential applications extending beyond agriculture to the general use of ultrasonic devices for pest control.