EOKHUB Faces

Pedro Basto

Summary

This interview with Pedro Basto, Program Manager for Research & Innovation at GICHD, underscores AI’s transformative potential for mine action, emphasizing tools like data management, ordnance recognition, and generative AI chatbots tailored to demining. A key takeaway from the recent Innovation Session is the need for structured and unstructured data—specifically, datasets rich in field-specific content to enhance AI’s applicability.

Basto highlights GICHD's commitment to ethical, responsible AI deployment, noting the importance of frameworks like the EORR Innovation Hub for fostering collaboration among governments, tech companies, and humanitarian organizations. With a strong background in EOD and C-IED from military to humanitarian contexts, Basto’s hands-on experience brings critical, field-driven insight into aligning technological innovation with deminers' real-world needs.

Interview

Pedro Basto, thank you for joining us. Most of the people in the explosive ordnance sector got to know you recently as Programme Manager for Research & Innovation at the Geneva Internacional Center for Humanitarian Demining (GICHD), but you have been working in the field of explosive ordnance for many years trying to bridge the crucial gap between the research and technology communities and practicitioners and fieldable applications, pursuing innovation that is meaningful and that adds value solving real needs-based problems.

Before we hear a little more about you, I would like to explore a little bit further this recent GICHD Innovation brand.

We are witnessing rapid technological advancements that are increasingly present in the field. GICHD has been always active in identifying, analysing and disseminating novel methods and technologies within humanitarian demining. Following up to the Explosive Ordnance Risk Reduction (EORR) Innovation Conference held in Nov23, just two weeks ago, GICHD organized a very successful global Innovation Session for the Mine Action community, on the subject of Artificial Intelligence (AI).

Why this theme and how do you see the role of Artificial Intelligence and machine learning evolving in EOD and mine action?

AI is a type of computer technology that allows digital systems to perform tasks that normally require human intelligence. Tasks such as understanding language, assisting the production of written records, recognizing images, data analysis, or generating decision-making options for complex problems. AI will do these things so well that the results will seem almost human-like in their accuracy.

There is a wide scientific consensus that AI will produce drastic changes in all areas of society, particularly because of the exponential acceleration of its pace of development in recent years, and even more so because AI is what the scientific community defines as general-purpose technology. Such technologies hold an immense transformative potential, because of they can be used in all fields of human societies. Past examples include the use of fire, electricity or digital technologies. When we look around ourselves, it is difficult to find a man-made object or activity that does not require or use those technologies.

AI will be widely used like electicity today. If this is the case, the sector needs to discuss when and how AI will became part of the deminer’s toolbox.

At the same time, while we have started hearing many voices within the sector showcasing how their tools and processes use AI, we have noticed that there is little common understanding of what exactly AI is and what it does.

Looking for answers for questions like, what is AI?, when and how can we use it?, was at the heart of our decision to do this Innovation Session on the subject of AI. We wanted to conduct a very dynamic event where we could raise the common understanding about the technology, contextualize it within the existing mine action frameworks of land release, information management and explosive ordnance risk education (EORE) and understand how can the sector be better prepared to harness the transformative potential of AI today or when such opportunity displays in the future.

GICHD Innovation Session on AI for Mine Action, October 2024 ©Antoine Tardy/GICHD

What have we learned from this Innovation Session?

I honestly think all involved learned a lot and it’s difficult to summarize the full dimension of the multiple interactions occuring thoughout the event. It was a learning journey for all participants.

Having to highlight some key lines to take, I would say acquiring or accesssing the right sets of data is an underlying requirement for any efficient AI application. There are multiples ways of tailoring the computational algorythm models behind AI, whether machine learning, neural networks, natural language processsing or others, but the data is the true raw material required. Both to teach and fine tune the algorythm models and to extract valuable results.

Today data doesn’t mean only structured data such as IMSMA forms, geospacial information, maps and other tradicionally associated sources of data. AI has also unlocked the possibility of resourting to vast pools of unstructured data such as any digital documents format, images, videos, social media among others.

For the mine action sector to be ready to use AI tomorrow, one needs to think today about the architectures of data collection, storage and sharing.

Another important takeaway was that despite the apparently extraordinary potencial of AI tools such as those deployed with geospacial AI applications or generative AI chatbots, all of these remain generic and somewhat blind to context.

We know from dozens of years of experience in humanitarian demining, that specific context, at regional, national and local levels, as well as, at technical level, is a key feature of our response to the risks posed by explosive ordnance to communities. If on one hand this represents a important limitation, on the other it’s an opportunity. One of the charateristics of AI technologies is its adptability to different applications and contexts, one of the reasons its considered a general purpose technology.

Did you attend the Innovation Session?

Yes
No
I am not interested.

What actionable outcomes could be shared?

The following two practical examples arising from the multidisciplinary thematic working sessions held during the second day of the Innovation Session tie together well with the points I have just made:

To be able to tailor generative AI chatbots to specific contexts, the first step would need to be developing and maintainng large language models that are sensitive to the context of explosive ordnance risk reduction. Teaching mine action language to chatbots if we would want to put it in simple terms. To do that, we would need as well to select and feed the system with the adequate datasets and ensure some form of quality control of the information fed into the AI model. Such models could make the dissemination of knowledge within the mine action sector extremely fast and efficient as well as accessible automatically in multiple languages. These models would allow the development of multiple support tools for revision and drafting of international and national standards, searching through information management systems, grants management and reporting, quality assurance of documentation and reporting, among many others.
To develop effective and efficient explosive ordnance image recognition applications, whether for supporting remotely operated survey and detection systems, early warning and protection of heavy equipment (mechanical demining machines, rubble clearance equipment, agricultural equipment or robotics), a sound reference database of explosive ordnance would be required to train effectively AI algorythm models. Such database could start with high resultion 3D models of explosive ordance. Enriching this initial database with imagery collected from the field and relevant technical information would allow it to be tailored to the specific local or technical context.

We worked on other 6 different themes, for each of them, drawing on the collective knowledge of multidisciplinary teams with representatives from mine action organizations working in the field, national authorities, international organizations, academia and research and technology organizations. Each thematic group devised a specific problem statement, key requirements for a solution and possible roadmap for development.

What is the most promising new technology in HMA?

Machine learning/AI.
Robotic systems.
Nothing mention

What measures are in place to ensure that the use of such technologies in demining does not inadvertently cause harm or displacement?

The ethical considerations and responsible use of AI is indeed a very important subject and one that was addressed as well. It’s a field of growing discussion within the broader technological and political and social sciences community. Although there are no global specific frameworks to enforce these ethical considerations in place yet, there is an emergent landscape of AI governance that give us very useful guidance, namely, the Montreal Declaration for Responsible AI (2018), EU AI Act (2024) and the UN General Assembly Resolution A/78/L.49 (2024) “Seizing the opportunities of safe, secure and trustworthy artificial intelligence systems for sustainable development”.

These framewords provide us with a few baseline principles for adoption in the design, development and maintenance of any AI applications: Dignity; Privacy; Human oversight; Transparency and explainability; Non-discrimination and equal treatment; Safety, Accuracy and Reliability; Understanding risk. These principles are grounded in the ethical background provided by the Universal Declaration of Human Rights.

We must be clear that any technology can be used for good or for bad purposes, that is a human decision that remains anchored with its designer, programmer or user. Even if humanitarian demining is an accepted good use for AI, the risks on unintended bad consequences must be accounted for, as with the application of any other technology.

Adopting sound risk management approaches from the early stages of development and adopting the principles outlined above are currently considered good practice to mitigate these risks and are likely to become mandatory, as it is the case under the existing AI governance frameworks within their jurisdictions.

GICHD Innovation Session on AI for Mine Action, October 2024 ©Antoine Tardy/GICHD

“Last year GICHD organized an wider event, the Innovation Conference. This year an Innovation Session on a specific topic. ”

You mentioned there are some key challenges hindering innovation. What are some of challenges of integrating new technologies into demining operations?

The key challenges we have identified hindering the uptake and meaningful impact of innovation in the mine action sector include the pursuit of “silver bullet” solutions, lack of needs-based approaches and appropriate legal and regulatory frameworks, cost-effectiness and limited funding, and resistance to change. These shortfalls result in redundant efforts, fragmented strategies, and an inefficient allocation of resources to innovation.

Moreover, lack of collaboration and coordination remains as well as a key challenge. This means, more specifically, that collaboration among stakeholders, including governments, international organizations, non-governmental organizations, academia, research and technology organizations, and local communities, is considered insufficient. If it's fair to acknowledge that all the challenges listed above are somewhat interconnected with each other, however we believe that collaboration and coordination is the crucial element that could help address all the others above, as well as those which I am sure we haven't uncovered yet.

This is the the problem statement underpining the concept and the establishment of the EORR Innovation Hub.

On a more positive note, despite the challenges, I would like to acknowledge the wealth of creative and innovative initiatives the sector has tried and learned from in the past. It’s that driving force to face the challenges in their specific context that we are trying to leverage. From the sector and for the sector is one of our core principles.

GICHD Innovation Session on AI for Mine Action, October 2024 ©Antoine Tardy/GICHD_

In practice, how is the Innovation Hub promoting collaboration and partnerships between mine action organisations, governments, and tech companies to advance technological solutions in demining?

Coordination and collaboration at the level of a global, multinational innovation system, such as the one we have within Mine Action or Ammunition Management, has no “silver bullet” either.

An innovation system is by definition a human system, complex, dynamic and composed with inumerous building blocks and their relantionships – in some cases collaborative, in other instances competitive, where different stakeholders have specific objectives, which are not always aligned.

First and foremost, this nature must be accepted. This means that for each case, we will need to look for the best tailored answer we can find, try it and be ready to readjust it at any time. Dealing with such uncertainty about the interventions and their expected results is part of it.

Trust, communication and understanding are some basic elements that significantly enhance the ability to navigate this system and improve the quality and quantity of its interactions – that is in fact what coordination and collaboration means from a complex systems approach.

We can build trust by bringing the different stakeholders together in open and voluntary forums such as the Innovation Hub, by asking the stakeholders to contribute to definition of the key topics discussed at the Innovation Conference, by bringing multidisciplinary teams to work together and brainstorm in safe environments such as in the Innovation Sessions, among others. This requires above all time, transparency and the abilty to broker the different connections within the system.

Finally, collaboratively establishing and describing the needs and requirements for research and innovation in the sector, through research agendas that gather a broad acceptance by the different stakeholders, will likely make it easier to draw additional resources for research and innovation. This is as well a relevant motivation for a wider collaboration and coordination.

You are part of?

National Mine Action Authority/Government
Mine Action Operator
Tech provider
Academia or Research Organization

Pedro Basto

Background

Could you elaborate on your background and previous experiences before joining the GICHD and what sparks your interest in the intersection of innovation and demining?

My first real contact with humanitarian demining happened along the demarcation line between the South of Lebanon and Israel, when I was deployed there as a UN peacekeeper in 2009. I was leading a military engineering unit of the Portuguese Army, in charge of opening patrol roads along the blue line and supporting the teams installing the blue barrels of the demarcation line.

Both tasks would be done after the clearance of hazardous areas which in most places was being done by humanitarian operators. Understanding exactly how that work was been done became a pressing need to perform my work, while at the same time, fed my curious mind. I had already done my combat engineer and sapper training, so it was with great enthusiam that I learned how humanitarian demining was done.

Hardly would I know that few years later I would embark in my EOD and C-IED journey, to become the lead of this operational capacity within the Portuguese Army for many years. During this period I gained the possibility of travelling all over Europe and a little of North Africa to share, learn and discuss the capacity development of EOD and C-IED capabilities. Either bilaterally, within NATO or within the European Defence Agency, I learned a lot.

In a parallel track, I started doing research related to engineering and explosives, such as implosion demolitions and blast resistance as early as 2007, coming in contact with academia, research and technology organizations and commercial partners.

In 2020, I joined the European Defence Agency to take up the position of Project Manager for the C-IED portfolio. If I had came across the challenges of bridging the gap between research and technology and the operational application of such innovative ideas before, it was during this time that I fully realized that I wanted to dedicate the next years of my carreer trying to fix that problem in a multinational setting.

And that’s when the extraordinary opportunity displayed, to do able to do it in a field where I had the technical expertize and with which I had such a personal rapport. Being able to contribute to humanitarian demining supporting innovation is a great honor for me.

I have a Master’s degree in Military Engineering, I am chartered Civil Engineer with an unfinished PhD on Blast Resistance of Infrastructures. Did 22 years of active service as a soldier of the Portuguese Army.

Could you give us a more an in-depth overview of your role at GICHD?

I currently hold the position of Programme Manager, Research and Innovation, within the Standards and Operations Division at GICHD. In this role, I provide direction to the Research and Innovation Programme, and the rest of the GICHD, focusing on leading and coordinating research and innovation related initiatives, and on methods that improve the delivery of explosive ordnance risk reduction operations. The aim of the Research and Innovation Programme is to support the identification, research and analysis of innovative methods and technologies, promoting their further development through partnerships, dialogue and cooperation and sharing good practice through studies, publications and conferences. We do this under the umbrella of the concept of the Innovation Hub, for which we act as secretariat and main coordination element, internally and externally.

It is a relatively new programme, with an ambitious vision and lot of work to do. But it’s also a young and highly motivated team, within an extraordinary open and collaborative environment across the GICHD. As said above, being at the intersection of innovation and the real challenges reaching us from the field is extremely interesting. It provides us the opportunity to work across various areas of expertize, from operational effeciency, to information management, EORE, standards or policy. In the end, innovation only makes sense when at the service of those areas. This means that every day there is something new to learn and incorporate in our work.

At the same time, it is a role which requires as well the ability to challenge how we do things today and helping others making and answering new questions, which is something I have done throughout all my career.

On a personal note, how do you relieve stress and stay fit and sharp? Do you have any hobbies?

I am interested by many and varied things in life. But what I think systematically helps me disconnect is reading, listening to music and riding my adventure motorcyle. I am an inveterated reader always with a pile of books around me. Either technical reading, history, graphic novels or classics of literature, reading feeds my constant desire to learn and understand what surrounds me. I am listener of heavy music. I think I can’t find an explanation for that – it’s just something that simply synchs with my heartbeat and resonates with me since I can remenber. And riding my motorcycle or travelling across the unbeaten track is just another way of seeking to learn about the world, I think. And of course, spending time with my wife, children and friends, with adundance of food and beveradges, is a must to keep the balance.