A New And Amazing Type Of Scarecrows Is Now Here – Autonomous Drones

According to researchers, flocks of pigeons on the roof of buildings have posed health hazards. Serious actions are now being taken by deploying some techniques of Convolutional Neural Networks to detect the location of pigeons on the roof of buildings and they are made to fly away through the autonomous drones. One such effort has been made by the team at Switzerland’s EPFL. Many of us like pigeons but they can create quite a mess on buildings and other structures. Research says that autonomous drones can be the best fit for harmlessly chasing them away.

According to a recent study, a team at Switzerland’s EPFL research institute has started installing a weatherproof pan-tilt-zoom video camera on the roof of the EPFL SwissTech Convention Center. The building attracts the folks of pigeons which cover the roof with droppings that need to be washed off.

By using a camera on the roof, the staying time of pigeons on the roof was observed within the time period of 21 days. Then by the usage of Neural Networks running on a linked ground station computer, the camera was able to detect the exact location of the pigeons in terms of GPS coordinates through the trained datasets. After the time period of three weeks, a drone, Parrot Anafi quadcopter, was also added to the mix.

Over the time period of five days, whenever the camera marked the pigeons on the roof, it was made to transfer the information to the drone. Then the drone took off, autonomously flying to the congested area, thus chasing the birds away. A human operator needed to authorize each takeoff for safety purposes. Over the five-day period, the drone was deployed about 55 times. As a result of this conduct, the amount of time that pigeons stayed on the roof was significantly reduced.

This system is described in a research paper, recently published in the journal IEEE Explore. The process can be seen in this video:

Leave a Reply

Your email address will not be published. Required fields are marked *