New millimetre wave radar can identify small drones with 99.9% accuracy

Advances in artificial intelligence, sensor technology and automation have propelled drones into mainstream use. But as they become more readily available, concerns are mounting over how to track them effectively, preventing them from posing safety risks.

This is where DroneTector hopes to come in: the startup is developing a millimetre wave radar to detect small drones at higher resolutions and rapidly quantify any threats.

“The application of this technology to the drone problem has been historically limited to research institutions,” says CEO and founder Matthew Moore, who built the company from his PhD research in physics at the University of St Andrews. “We’re taking it and turning it into a viable tool for industry.”

Traditionally, high-frequency radar systems relied on complex physics and advanced materials, making them impractical for large-scale commercial deployment. In recent years, the development of autonomous cars has led to key advances in millimetre wave technology at the semiconductor level, to enable vehicles to map their surroundings for navigation.

Moore has combined these new components to create a powerful radar system that’s scalable and more precise than ever.

“Our innovation lies in how we take these building blocks — components that weren’t previously available — and package them into a system that works at scale,” Moore says.

On the software side, following dramatic improvements in neural network-based classification algorithms, the company has developed sophisticated, yet lightweight, machine learning models that can run on a portable platform and can track and classify large numbers of drones faster and with 99.9% accuracy.

“Speed is critical,” Moore says. “You could have hundreds of targets in your field of view at any given time, and traditional classification methods couldn’t handle that volume quickly or effectively. Also, we’ve been pushing the limits of algorithm accuracy because that’s the gold standard here. In almost all situations, you’re going to see many more birds than drones, so if you have an algorithm that has 99% accuracy, it may sound good, but you’re actually getting some false detections.”

Millimetre wave technology uses shorter wavelengths to produce high-frequency signals and detect echoes, capturing higher-resolution signatures from drones. Compared to existing lower-frequency radars, this allows DroneTector to accurately locate drones in all weather conditions and at night, including fully autonomous drones, while also differentiating between models to quantify the type of threat.

For Moore, in order to fully leverage this technology, it’s crucial to predict and address the potential consequences of its misuse.

“We’re clearly ill-prepared,” he says. “It’s really too easy to get a hold of this stuff — you can buy it on Amazon — and it seems to me like a nightmare situation: you have these small invisible objects that can fly around and cause untold damage without really having a sense of whether they’re coming or how to protect yourself. And when you combine this with AI…”

Back in 2018, Gatwick airport shut down for 36 hours to investigate a series of suspected drone sightings close to the runway. The incident cost the airport £1.4 million, while airlines faced around £50 million in compensation to passengers and lost revenue.

Though similar large-scale incidents are still rare, airport disruptions are becoming more frequent as the use of drones proliferates, and solutions like DroneTector could help minimise their impact.

“Our system would be scanning the air at all times, so we would see drones before any sightings,” Moore says. “But let’s say there needs to be a secondary check — we can survey that specific area, and within a few seconds airport operators would get an answer, so there wouldn’t be any real impact to flights.”

The company has recently secured a pilot with an airport to test the technology, and is exploring potential dual use applications.

About DroneTector and Conception X

DroneTector was incubated at Conception X. Moore joined Cohort 6 to commercialise his research while completing his PhD.

“As a PhD founder, you can feel like you’re too junior, too young for something like this — that maybe it’s not a realistic goal,” he says. “It’s not like you’re surrounded by examples of people in your day-to-day life who’ve done it. Conception X changes that, and makes it seem possible.”

“My entire commercialisation network comes from Conception X. I was surrounded by a cohort of a hundred people who were facing the same challenges, which makes all the difference when you’re trying to commercialise your PhD, especially when it often feels like no one else around you is doing the same.”

DroneTector is currently raising a pre-seed round. Get in touch with Matthew Moore to learn more.