Ilija Mišković – Using AI And Machine Learning To Create Sustainable Solutions

Ilija Miskovic

My work in AI and machine learning represents the culmination of years spent at the intersection of science, technology, engineering and mathematics, striving to create more efficient and sustainable solutions for our society.”

What led you to engineering, and to your current work in AI and machine learning?

At the age of 19, I started a challenging but very rewarding journey, working full-time as an electrical technician while simultaneously attending university and completing my first degree.

My focus during that period was on process and energy systems, which included participating as a principal designer, consultant and project manager on over 100 projects in power generation, transmission and distribution sector.

I loved the work and continued on doing a master’s degree, this time with a focus on process systems in environmental engineering. My master’s thesis looked at recycling waste electronic and electrical equipment. It connected to the work I was doing – for example, when you have large transformers at the end of their useful lives, how do you recycle them?

In 2007, I moved to the US to do a PhD in the mining and minerals engineering department at Virginia Tech doing research on carbon capture and storage.

During my time at Virginia Tech, I was part of some very exciting projects. Our team secured significant funding from the US Department of Energy to deploy geological carbon sequestration – we successfully completed one of the first field validation studies of geological carbon sequestration in unmineable coal seams.

My journey through various fields of science and engineering gradually led me to the fields of machine learning and AI.

The transition was a natural progression, born from a desire to harness cutting-edge technology to solve complex challenges associated with energy production and extraction of natural resources. Today, my work in AI and machine learning represents the culmination of years spent at the intersection of science, technology, engineering and mathematics, striving to create more efficient and sustainable solutions for our society.


Tell us about your research.

I cannot claim to be exclusively a mining engineer or a process engineer. What I am is an applied mathematician. I understand real-world engineering problems and describe them through equations, and I then develop models and laboratory- and pilot-scale experimental validations of these concepts so we can scale them up further and ultimately apply them in the industry.

My research gives me a very diversified exposure to a range of problems. But it always comes back to the first principles of science and mathematical models.

My current research is far-ranging, and includes such areas as AI-assisted geoscience and resource modelling; digital twins and smart proxies; autonomous cyber-physical systems for remote operations in extreme environments; industrial microgrids; edge computing and embedded AI; federated and distributed learning; industrial 5G NR technologies and networks; and real-time data analysis and visualization of large multidimensional data sets.

Much of my current research is in cyber-physical systems for remote operations. I am looking at how to coordinate distributed operational assets (equipment and personnel), in remote and challenging environments.

I was the first researcher in North America to develop a 5G-enabled mine test bed.

This project involved using a fleet of scaled down hydraulic mining trucks equipped with different exteroceptive and interoceptive sensors to simulate different use-cases for application of 5G networks, edge and cloud computing, and distributed and federated learning within the context of Level 4/5 autonomy – one that enables remote assets to operate individually without the need to communicate with a central agent.

I am also using machine learning to expedite the task of discovering mineral deposits and assessing their mining potential.

My students and I are using different types of geological and historic (mine) production data, which we use as various input layers to our machine learning algorithms, to identify where the occurrence is most probable for specific minerals. We recently completed a mineral prospectivity mapping study for a 300 square mile gold exploration area in northern British Columbia, using only our machine learning algorithms and publicly available geological and production data.

This study, funded by an exploration company, was aimed at validating the utility of our data-driven models in real-world applications.

Without prior knowledge of the results and maps from the company’s own field exploration and drilling programs, our models successfully identified a number of prospective locations within the exploration area, with an overlap of more than 90% with what they had discovered through their exploration programs, but at a many-fold lower cost.



Why is this research important?

Production of minerals and metals is a critical step in the transition to a more sustainable economy.

Finding more efficient, faster and sustainable ways to discover and extract these commodities can accelerate that transition.

My work on using AI to locate areas of high mineral potential can speed up the process from discovery to production. Many mining companies spend upwards of 10-15 years to discover a viable deposit, which also comes at great costs. This work could help expedite the process, which would then allow us to get the minerals we need to develop more affordable green technologies much faster.

The research on autonomous cyber-physical systems has a lot of applications, given that mine sites are often located in remote and extreme locations.

We already see this in Australia where there are operators at control centres in large cities like Perth, controlling remote trucks at mine sites thousands of kilometres away in the outback. My work takes this one step further by using advanced 5G networks and AI, which will enable the trucks to operate more autonomously and with improved privacy and security.


What undergraduate courses do you teach?

I am teaching courses on control and automation, and on expert systems in mining and minerals processing. I also teach geostatistics. I am always updating my teaching and materials based on what is current in the field – I try and bridge the gap between the theory we are teaching in the classroom with practical examples from my research or with industry-relevant use-cases.

Why should students consider mining engineering?

Mining engineering is such a broad field that it’s possible to take your career in many different directions.

As a student, you not only learn the foundations in the domain of mining but also delve into areas you might not immediately associate with mining, such as computer science, programming, automation and control, advanced energy systems and robotics, among many others. Our graduates find themselves well-prepared for a wide range of career paths, including opportunities in consulting, financial institutions, government, the IT sector and beyond, showcasing the versatility and demand for skills acquired in this field.


Why should students choose UBC?

I have studied and worked at many different universities in Europe and the United States, and, based on my experience, UBC stands out as an exceptional institution.

The faculty across various departments are top-notch, bringing a wealth of knowledge and experience to the classroom.

Also, students have a wide range of elective choices, allowing them to enhance their educational experience. There is also excellent industry support for our programs, providing students with numerous opportunities to work on real-world problems. This combination of factors makes UBC an ideal place for academic and professional development.