The United States government is looking to develop urban drone dragnet systems capable of identifying Unmanned Aerial Systems (UASs) that are usually hard to detect in complex environments.

The system proposed will involve a network of high-level sensor platforms, either tethered or using long-term endurance hovering systems, which will work in cooperation above cities to provide authorised personnel with a Common Operational Picture (COP) of sub-flight airspace, providing identification systems which can operate beneath the margins of conventional aircraft surveillance systems.

The request includes an abstract representation of the scope and aim of the system, which seems to be employing multiple viewpoints to assemble a kind of holographic-style identification of UAVs which no single viewpoint could establish:

DARPA-BAA-16-55-Aerial-Dragnet

‘Small UASs are rapidly becoming low-cost aerial platforms for hostile reconnaissance, targeting, and weapon delivery. Unlike traditional air targets, small UASs: 1) fly at low altitudes (e.g., < 400 ft) which make them easily hidden by complex terrain, 2) move at slow speeds (e.g., < 90 kts) which make them difficult to differentiate from other movers, and 3) are small in size (e.g., < 55 lbs.) making them difficult to sense. In future urban battlegrounds, U.S. forces will be placed at risk by small UAS which use buildings and naturally-occurring motion of the clutter to make surveillance impractical using current approaches. The rapid proliferation of commercial UAS with increasing endurance and payload capacity drives the need for a future urban aerial surveillance system that can detect, track, and classify many different UAS types at longer ranges in urban terrain’

One reason that a multiple-node system is necessary for a comprehensive UAS identification system is literally regarding perspective. It is difficult to estimate the elevation of a discarded carrier bag eddying in the wind without at least a stereo view of the environment – triangulation can confirm where a tiny, moving object is in relation to the ground.

The request notionally limits the cost of covering 20 square kilometres to $10,000, anticipating a 9-node city-wide network to be delivered for less than $90,000.

The Aerial Dragnet program will take place over the next three and a half years and the final system must be capable of identifying target drones within 10 seconds of detection, with a 95% rate of accuracy and a false target rate of < 1 per 24 hour period.

Additionally, the paper notes the need for multiple types of detection system, possibly including radar and advanced optical evaluation systems, for redundancy and confirmation. Since the objects the system will be searching for can be so small, the paper suggests that a system may include a method of detecting the micro-Doppler signature of an airborne object; birds as well as UASs can generate these signatures, so the ambit is to reduce false positives as well as take advantage of any residual technological signature drones may still provide as they evolve.

Though the request states that no proposed solutions simply be evolutionary extensions of existing systems, it also requires solutions that employ off-the-shelf or previously developed components to keep costs down.

The invitation is limited to contractors and individuals native to the U.S., and DARPA is holding a Proposer’s Day conference for the project on September 26th at Arlington, to provide additional information and address enquiries from potential proposers.