The ever expanding pressure to fulfil the planets food demand as the planet’s population is increasing year by year. Computing technology has aided smart farming from self driving tractors, autonomous drones to monitoring crops for insect infestation. One of the areas of smart agriculture is smart trap fruit fly monitoring, if we can detect certain species prior to being picked and delivered, farmers can take preventive action. One of the biggest challenges of such a fruit fly monitoring solution is being energy aware.
(M2M) Smart traps for fruit fly do exists in the market like (Trapview) but don’t take advantage of edge/fog computing and AI is performed in the cloud. This thesis demonstrates that if we observe and record the fruit fly’s behaviour along with deep learning we can introduce a novel method of AI energy aware.
This is achieved by creating a smart trap made from Acrylonitrile Butadiene Styrene (ABS), with a pheromone attractant to lure the fruit fly’s intro the smart trap. We implement a embedded solution with photo optics and environmental sensors that count and timestamp the insect enters along with the interval between each fruit fly and its environmental surroundings. Lora wireless communications is used to transmit sensor data to fog computer for AI processing, this implementation will be tested on live farmers fruit crops throughout the UK this April.