Skeptics and Minefields: Breaking New Grounds in De-Mining Techniques

AuthorAuthorAuthor

Jannic Meyer, Jack Wang & Dmytro Titov
July 18, 2024

Welcome 🇪🇺

This week we are happy to have Dmytro Titov, contributing to the topic of de-mining that we touched in December of last year. He is sharing his first hand experiences about the Ukrainian mine situation, what impact technology could have for de-mining and which challenges are faced on a day-to-day basis.

The author is the founder and CEO of Ailand Systems, a company that develops autonomous drones. The company pivoted from agricultural drone technology to de-mining, addressing the urgent need to clear vast areas of land contaminated by mines, starting in Ukraine.

Yours,
Uwe, Jack and Jannic

Security & DefenseTech Event in Switzerland on August 15 🇨🇭

After a successful launch event in February, a second ‘K-Works’ will be held bringing together entrepreneurs in the security & defense sector in Switzerland. The August 15 event, hosted by Deloitte in Zurich, Switzerland, will try to shed light on the challenges small technology companies face entering larger foreign government markets.

The evening will be concluded with an apéritif’ to network with a diverse group of entrepreneurs and leaders from technology, military, defense procurement, venture capital, and academia, seeking to improve the interplay of these sectors.

[Link] to more information & registration

For questions contact Rico Chandra or Twitter/X @unkenruf ]

The Resilience Conference 🇪🇺🇬🇧

We are excited to be supporting the Resilience Conference in London in 26 - 27th of September. You can find a full list of amazing speakers here including Mike Butcher from TechCrunch, Blythe Crawford from Air and Space Warfare Center, UK Royal Airforce as well as GPs from funds like General Catalyst, DCVC and of course our very own Uwe Horstmann.

Learn more and apply for a ticket: https://www.resilienceconference.io/

Breaching through the minefield of skepticism

Ukraine is the most mine-contaminated country in the world. The government estimates the total potentially contaminated area as 174K km², which is about 1/3 of the country's total area and can be compared to the size of Great Britain.

The World Bank estimates the total cost of demining in Ukraine to be 37 billion dollars. The process will take decades and, unfortunately, many human lives.

We need to find a way to make this process faster, cheaper, and safer. That’s exactly where autonomous unmanned solutions can help. Many teams, including my team, are tackling this problem, and I’d like to share two main challenges that we currently face.

Technical challenges

Landmines can be detected using various sensors: metal detectors, magnetometers, ground-penetrating radar, etc. Each sensor has pros and cons, but the biggest issue common for all sensors is false positives rate: how often detected objects will turn out to be clutter, stone or phantom signals caused by high ground mineralization or electromagnetic noise. Sometimes false positives can constitute 80% of all detections, which deems results useless. If we tune signal processing thresholds to get rid of false positives, then we start missing real objects, i.e. we start getting false negatives, which is even more dangerous. Finding optimal balance between the two is an art that only human sappers could master. At least till now. Let’s see how they do it.

The main tools used by sappers for landmine detection are eyes and metal detectors. Eyes "are used" to detect visible objects, like scatterable anti-personnel landmines, unburied anti-vehicle landmines, unexploded ordnance, etc. Metal detectors are used for sub-surface landmine detection. A good metal detector in the hands of a good sapper can detect even the most challenging plastic landmines with low metal content, like PMN-2 or TM-62P3. To find those mines, sappers perform systematic metal detector swipes from side to side right above the ground, waiting for an audio signal from the detector.

Eyes and metal detectors are an industry standard for humanitarian landmine detection.

If we look at this process from the perspective of a robotics engineer, it’s not something fundamentally difficult to automate. Swipes are trivial movements, which does not require advanced 6DOF mechanics - it can be implemented using a single servo motor and distance sensor. As for signal processing, it is already implemented as part of metal detector software - the sapper only waits for the result of the processing in the form of a beep sound. What makes sapper really important is the decision taking part: is this signal from a real landmine, or is it clutter? Should I try to scan this area with different speeds, frequencies, different search coils? This decision is where the main technical challenge lies and that's a great application for machine learning approaches or broadly speaking AI. We need to train our model based on right and wrong decisions until it is able to continue making the right decisions on its own, achieving human-level accuracy.

So AI is the answer, and if you ask why it has not been implemented yet - that’s because technology has matured enough in the current decade. Only now can we put enough computing power and energy onto a small autonomous platform to make real-time signal processing and decision-making sufficient? Besides hardware, AI models have also advanced greatly during the last few years, as we all could see. That’s why my team and I personally believe full-scale solutions for autonomous landmine detection will emerge in the coming years. It still takes much time to put everything together, conduct all necessary tests, and assert accuracy in all kinds of environments, but the process is ongoing and is unstoppable.

Justified skepticism

The second challenge that we face is skepticism that is coming from demining operators. According to Ukraine laws, a demining operator is a company certified for humanitarian landmine clearing operations. Such companies are the main consumers of potential autonomous solutions that can significantly scale up their operations. Ironically, most of the companies are skeptical about using drones. And there is a reason for it. They saw too many solutions that failed. Unfortunately, multiple teams claimed too early about solving the landmine detection problem, offering their solutions as a final product and claiming high accuracy. However, all those solutions have failed miserably and couldn’t compete with human sappers. In fact, they slowed down the operations due to high false positive rates that we discussed earlier.

Some of the solutions we saw cost hundreds of thousands of dollars and couldn’t detect a single landmine. Unfortunately, there is a tendency to skip necessary tests and deploy solutions as early as possible and many companies exploit this opportunity in their favor.

Now, it’s going to be harder to penetrate the market even with a working solution. Someone has to break through the minefield of skepticism. The only way to do it is to avoid rushing, do all necessary tests in different environments and soils, and get certified by a third-party authority. It will take time and investment.

But that would be an impactful investment. In addition to the usual financial returns, ROI would also include the amount of land released to production and the amount of lives saved.

News That Caught Our Attention 👀

Every week we feature a list of interesting roles in European DefenceTech start-ups and scale-ups for readers seeking their next challenge in their careers.

If you are a founder and would like to promote your open roles, please get in touch with us!

Passionate and want to contribute? 👩🏻‍💻

The European Resilience Tech Newsletter is always looking for regular and guest authors, writers, reporters, content creators etc. If you like what you read, you are passionate about improving European resilience regardless of your background and want to contribute, just reach out to us!

European Resilience Tech Newsletter Team

Uwe Horstmannco-founded Project A Ventures in 2012 as General Partner and has built Project A to be a leading European early stage investor with over $1bn USD under management and having backed 100+ founders. In addition to Project A, Uwe serves as Reserve Officer in the German armed forces and advises the German Ministry of Defence in digital transformation issues.

Jack Wangis a software engineer turned product driven tech investor and joined Project A in 2021 to lead the firm’s deep tech investing, which have grown to include DefenceTech. Prior to joining Project A, Jack worked in a variety of organisations such as Amazon and Macquarie Group across Australia, US and UK / Europe. Jack holds a MBA from London Business School and Bachelors of Engineering (Bioinformatics, 1st) from UNSW, Australia.

Jannic Meyer joined Project A initially contributing to what is now known as the Project A Studio, partnering with founders at the pre-idea stage, where he covered a variety of topics ranging from energy infrastructure to dual-use robotics and led our investment in ARX Robotics. He is now part of the investment team at Project A covering all things resilience.

Project A Ventures is one of the leading early-stage tech investors in Europe with offices in Berlin and London. In addition to 1 billion USD assets under management, Project A supports its 100+ portfolio companies with a platform team over 140 functional experts in key areas such as software and product development, business intelligence, brand, design, marketing, sales and recruiting. Project A have backed founders of Trade Republic, WorldRemit, Sennder, KRY, Spryker, Catawiki, Unmind and Voi as well as founders building in European Resilience: