NOCTUS

Detect, identify and intercept drones for securing high value areas such as the prison, civilian buildings, etc. within a cost-effective package

In a world of Automation and Robotics, we as humans have all seen or thought about technology’s potential and it’s extreme outcomes. We’re so much invested in the future that we forget what is happening in the world around us. We see that new inventions created for particular cause end up in the wrong hands, as instruments to unwarranted destruction. One such creation is the Drone.

Although, the idea of a Drone was initially used for Military purpose by the Austrians in 1849, and the fact that it started off in the modern age as a weapon, we see drones being used innovatively now in commercial backgrounds and free public use. The increased accessibility to drone technology has led to an increase in terrorist usage and unsolicited intrusion.

Drones are now being fitted with the latest in Spying hardware and weaponry. We see Stealthy Drones which have infinitesimal RADAR signatures, some are sound suppressed, you name a feature and there is a Drone created with it. This increase in Tech accessibility, combined with ingenuity, pose a very big threat to Civilian Infrastructure where innocent people thrive.

On the left, A downloaded drone image via search engine. Right, the inference rendered output after passing through object detection pipeline.

A Solution is required to tackle this problem in the civilian environment with affordability at it’s core and reliability as a key criterion. This Project aims to provide a concept or a working model of a system which can help Detect, Identify and Intervene with Drones where they are not expected to be.

demo of drone detection based on live video feed. inference running on ~30 FPS on a Ryzen 7 + RTX 2060 laptop

This work was done while consulting and leading a group of 10 students at the Aalto Univeristy, Finland as a part of Product Develepment Project funded by Saab AB. Here is the detailed report.