Mining and Automation – How an Industry is Addressing Challenges
Since hitting its lowest levels in 2016, the mining industry has witnessed rising demand from the renewables and IT sectors. Increasing capital expenditure and declining productivity levels has led to limited access to quality resources. Also, fluctuating demand (largely from renewable systems demand) and a lack of industry flexibility has led to market instability. Given these factors, emerging startups in mineral exploration are aiming to address these challenges, with a focus on enabling increased discovery rates and reduced costs.
Attractiveness
Approximately $3.6 trillion of capital investment will be required to find new resources between 2013-2030. Currently, the cost of exploration has increased exponentially, while the number of discoveries has reduced. Generally, there is a <1% chance of finding a deposit, with 10,000 prospects needed to be surveyed to find 1 feasible ore body.
Artificial Intelligence (AI) and machine learning can sift through large quantities of data, generating models to predict areas with similar features that have known deposits, reducing the amount of labor required. Drones and autonomous mapping devices allow visual access to remote and unreachable underground areas without endangering workers.
Recently, mineral exploration has attracted players from the technology industry such as Breakthrough Energy Ventures. With concerns rising about the source and footprint of raw materials, new investment streams are opening. The Australian government also launched an Accelerated Discovery Initiative that is focused on mineral exploration, backed with a fund totaling $10 million.
Another key enabler in the market is Unearthed Solutions, a for-profit company based in Australia, that connects corporates to research which can solve their challenges. Holly Bridgewater, industry lead at Unearthed Solutions, considers digital transformation as a positive for mineral exploration but states “the transformation should not be purely technological but incorporate the skills and value people bring to improve overall productivity”.
Emerging Business Models
Mapping and Analytics
 Exyn Technologies, founded in 2014, develops autonomous mapping software systems that, coupled with a DGI drones, map out 10 times more data points per second compared to conventional solutions. The solution is self-contained, does not require satellite navigation, external communication or piloting, and is sold as a complete product with support provided. Exyn builds and uses models for multiple sensors including Lidar and employs multi-sensor fusion.
It received $16 million in Series A funding in a round led by Centricus in 2019, bringing total investment since launch of the company to $23 million+. |
|
 Emesent, founded in 2018, has developed an autonomous mapping software, packaged in a modular device that can be handheld or attached to a drone. It allows for collision-free, autopilot mapping using Lidar, and is sold as a software-as-a-service with support provided. It is currently working on developing software that is compatible with drones other than DGI, giving it an edge on its competitors. It obtained seed funding of $2.5 million in 2018, led by Main Sequence Ventures. |
|
 GeoSlam, founded in 2012, is a provider of satellite navigation-free mapping software, that can be handheld or attached to a drone or vehicle. It allows for volume and tonnage calculation of mapped areas and is available for sale through vendors with support provided.
Its products are being used by Newrange Gold Corp, an exploration company, to map tunnels for sampling. It has also partnered with Normet, an equipment manufacturer, to develop combined software and hardware solutions. |
Machine Learning
 Imago, founded in 2016, applies machine learning to store, integrate and analyze geoscientific images to identify properties of rocks and generate insights. The solution is available as software-as-a-service with support. Imago can use images of any kind and size and analyze them as opposed to its competitors which analyze samples directly. Imago partners and integrates with other machine learning engines and exploration software to provide a complete solution. Imago is currently focused on mineral analysis but will venture into coal and the oil & gas sectors in the coming years. |
|
Earth AI , founded in 2016, uses satellite data fed machine learning to generate deposit probability maps. It employs several business models involving cash rewards or joint partnerships with mining companies, including buy in or out of an existing project, or initiating a joint exploration project. They currently have a 26% success rate of finding a deposit as compared to the industry standard of 5%. It received $2.5 million seed funding from Y Combinator and Gagarin Capital in 2019, and is part of Y Combinator’s 2019 cohort. |
|
 KoBold Metals, founded in 2018, is using AI and big data science solutions to find new ethical sources of cobalt, with transparency throughout its value chain, for use in the technology industry. It aims to collect and analyze data from satellite imagery and historic drilling results to generate a global deposit map of the earth’s crust. It has received seed funding of $20 million from Breakthrough Energy Ventures and Andreessen Horowitz in 2019. In 2019, energy company Equinor announced plans to take a stake in Kobold Metals, due to Equinor’s interest in applying AI to find new oil & gas deposits. |
|
 Goldspot Discoveries, a company founded in 2016 and publicly traded in 2019, is employing AI and machine learning to find deposits with its technology which is validated by clients such as Vale. It offers combined geological and AI solutions, with analytics-as-a-service for mineral exploration and insights-as-a-service for resource investing. It also partnered with junior miners such as Manitou Gold and Newfound Gold Corp to identify areas with greater probability of gold deposits, for royalties in the project. |
Emerging innovators are situated in areas with proven mining markets such as Canada, Australia and the USA, giving them rapid market access. To avoid direct competition with mining corporates, these startups are focusing on relatively stand-alone segments within the industry. For example, Earth AI sells the rights to its find for a stake in the project to a mining corporate rather than developing a mine itself. KoBold Metals is also expected to implement a similar model.
Competition and Challenges
Given the pressure on the mining industry to reduce its carbon footprint and the rise of ethical investing, investors are wary of funding the industry. This shifts corporate’s focus from innovation to retaining social license to operate and mitigating negative effects of their mining operations. For example, Glencore is using virtual reality to train miners, starting with its mine in Zambia in 2016, while BHP is incorporating the use of underground fleet management systems for task management, proximity detection, and worker safety starting 2019.
AI is being used for preliminary exploration as finding precise locations is still not achievable with the current state of the technology, therefore still requiring supporting geological surveys and multiple drilling processes. This can be improved by integrating data from multiple sources and systems and enhancing collaboration between mining corporates, junior miners, and data science innovators. AI-driven mineral exploration is still not mainstream, with a few corporates employing the technology very recently:
|
|
|
|
|
Conditions for Success
For start-ups’ AI solutions to succeed, their systems need to be fed insightful data to adequately train them. Goldspot Discoveries and Imago are partnering with junior miners to use their extensive geological and geographical data while Earth AI is using satellite data from NASA. Corporates do not want to invest in technology that is not proven, but the nature of the AI solutions requires access to a company’s mining data to render results. This presents a challenge to innovators, who need to generate credibility of their solutions before key partnerships can be made. Alternatively, innovators can use opensource geological data available through regional governments to train AI.
Other innovators are becoming aware of the opportunity in mining and are expanding to this sector. DroneDeploy and Microdrones are applying their unmanned aerial vehicles for exploration, and although their drones do not have extensive mining specifications, they do have better hardware and drone control. Koan Analytics, a data aggregation and analytics innovator, is applying its machine learning expertise to mineral exploration. Bryn Wyka, founder of Koan Analytics, highlights the challenge with using mining data to feed AI, “the real challenge is converting the unstructured mining data into structured data that AI can understand and use”. Koan Analytics is testing making pdf files readable by AI, which will allow mining companies to develop insights from their data easily.
Hurdles ahead
While AI is projected to increase productivity and reduce costs, there are concerns about its impact on social relations. Automation could take away 44% of unskilled workers jobs’ by the mid 2030s, drastically reducing the employment benefit mining operations provide to local communities and the economy. The skills gap within the industry is also hindering the adoption of advanced solutions as more of the current workforce possesses basic skills. The industry is not as attractive to younger tech-savvy people leading to a further increase in the gap.
And Finally…
Although AI is relatively new, its use and need in mineral exploration is evident given the increased discovery rates observed. With the projected growth and advancement of the technology, corporates need to increase their efforts to partner with emerging innovators to stay ahead of competitors and increase productivity.
