Editorial for the Special Issue “Mine Automation and New Technologies”

1. Introduction

Mining is undergoing a transformation driven by digitalisation and automation, promising improvements in efficiency, sustainability, and safety. Specifically, mine automation plays a key role in removing humans from hazardous environments, thereby reducing accidents and enabling the vision of “zero-entry” mining (i.e., no personnel exposure at the active face in hazardous locations). Automation helps cut emissions and supports greener, more sustainable operations. Meanwhile, the depletion of high-grade mineral resources prompts interest in unconventional mining methods, such as deep-sea mining, urban mining (metal recovery from waste), in situ leaching, and even space mining, to access new reserves as conventional ones dwindle and satisfy the high demand from different industry sectors [1,2]. Therefore, research is essential to assess the long-term stability and efficiency of novel artificial intelligence (AI)-powered technologies in harsh mining environments. Additionally, examining the potential for extracting alternative mineral resources and the challenges associated with them is of great importance.

This Special Issue of Mining (MDPI) on “Mine Automation and New Technologies” [3] showcases state-of-the-art research at the intersection of advanced technology and mining engineering. The papers cover a wide range of topics: AI and machine learning for solving complex mining problems, autonomous systems for mine operations, advanced sensing and simulation for monitoring and planning, and explorations of alternative mining approaches. Collectively, these contributions highlight how embracing digital transformation and novel techniques can enhance productivity, safety, and sustainability in the mining sector. The editorial concludes with a summary of cross-cutting insights and future research directions.

2. Machine Learning Accelerating Mining Optimisation

A prominent theme within this issue is the application of machine learning (ML) and AI to model and optimise various mining processes. Several contributions demonstrate how data-driven algorithms may yield more accurate predictions and efficient planning than traditional methods:

Predicting coal abrasiveness: Afrazi et al. (Contribution 1) used tree-based ML models (Random Forest, Gradient Boosting, and XGBoost) to predict the abrasive index of coal from its mineralogical properties. After cleaning the dataset and selecting key features, the best model (XGBoost) achieved a high accuracy (R² ≈ 0.92), substantially outperforming the initial models (R²~0.63–0.72 before refinement). The analysis identified quartz and ash content as the most influential factors increasing abrasiveness, while calorific value has least impact. This data-driven approach may offer a cost-effective alternative to laboratory tests, allowing industry to predict material wear characteristics in advance and optimise coal blending and maintenance schedules.

Optimising blast vibrations: another paper (Saubi et al., Contribution 2) addressed ground vibration control in open-pit blasting through an AI-based model. Using data from 100 blasts, the researchers trained an artificial neural network (ANN) to capture the complex relationship between eight blast design parameters (burden, spacing, hole depth, charge per delay, etc.) and the resulting vibration. By treating the trained ANN as a “solution surface” in a nine-dimensional space, it was possible to search for the combination of inputs that minimises vibration. The model suggested that, under ideal conditions, peak vibration could be reduced to around 0.1 mm/s—roughly equivalent to the ground motion from a passing vehicle. While such an extreme reduction may be theoretical, the study demonstrates how ML optimisation may guide blast designs to minimise environmental and social impacts.

Process modelling with limited data: a challenge in applying ML to mineral processing is the large data requirement. One innovative contribution (Moraga et al., Contribution 3) showed how to train neural networks with minimal data for a comminution (crushing/grinding) circuit. By carefully selecting only 77 representative input scenarios using a conjoint analysis approach, the authors trained neural nets to predict key process outputs (such as energy and water usage, particle size distribution) based on simulator-generated data. Despite the small training set, the networks achieved excellent predictive performance (R² > 0.99 on training scenarios) and maintained high accuracy (R² > 0.98) on new test scenarios within the considered range. The only limitation observed was a loss of accuracy for inputs outside the originally sampled range. This approach reduces the burden of data collection and shows that even complex process dynamics can be learned with limited but well-chosen data. It opens the door to deploying ML models in mineral processing plants where historical data are scarce, thus simplifying and improving process optimisation.

Optimising truck and loader fleets: efficient fleet management in mines is another area enhanced by ML. A case study (Nobahar et al., Contribution 4) from a Kaolin mining operation applied a gradient boosting algorithm to optimise the selection of trucks and loaders needed to meet daily production targets. Five years of operational data (weather, number/type of trucks and loaders, daily tonnage hauled, etc.) were used to train and validate several models, among which gradient boosting performed best (approximately 85% accuracy in prediction). The trained model was then fed over 11,000 hypothetical fleet scenarios to identify the optimal equipment combination for each part of the mine section and for each day. From this, the system recommended the configuration that met the required output with minimal idle time (e.g., specifying ideal truck numbers and loaders for each shift). This data-driven fleet optimisation promises to reduce truck queues and shovel downtime, cutting operational costs. It illustrates how ML can be leveraged as a decision-support tool in mine planning, enabling managers to explore numerous what-if scenarios quickly and choose the most efficient plan without costly trial-and-error in the field.

3. Automation and Human–Machine Collaboration in Mining

Mining equipment automation is another key focus of the Special Issue. Several papers explore the current capabilities and future possibilities of autonomous systems in mining, often stressing the importance of human integration with these technologies:

Human-in-the-loop autonomy: a forward-looking conceptual paper (Ruiz-del-Solar, Contribution 5) advocates for “autonomous collaborative mining” as a new paradigm for mine automation. While completely human-free, fully autonomous mines remain out of reach with today’s technology, this concept suggests that safety and efficiency can be maximised by having people and automated machines work together rather than separately. The authors note that many “autonomous” machines still require operators, and many tasks remain unautomated. Rather than viewing this as a setback, the authors propose deeper personnel involvement inside the automation loop: for example, remote operators could guide or intervene in autonomous operations when complex judgements are needed, people would make strategic decisions with AI support. This collaborative approach could accelerate progress toward zero-entry mining by leveraging human judgment where automation is weak and using robots where people would be at risk. The paper analyses enabling factors (such as advanced operator interfaces and communication systems) and presents a case study of an autonomous load–haul–dump machine supervised by people to demonstrate the concept. The broader implication is a shift in mindset: future mines may not be 100% automated, but rather symbiotic environments where operators and robotic efficiency complement each other and improve operations.

Autonomous shovels and productivity: a study by Yaghini et al. (Contribution 6) evaluated the impact of automation on an electric rope shovel used in surface mines. Using discrete-event simulation with real operational data, the researchers modelled four scenarios, ranging from a manual operation to fully autonomous operation, with intermediate levels such as operator-assisted movements. The findings were striking; even partial automation (e.g., automating the swing and dump motions) improved performance, and the fully autonomous scenario showed a potential 41% increase in output compared to the manual operation. These gains resulted from more consistent cycle times and reduced idling, demonstrating that automation can significantly boost productivity in load–haul systems. The study suggests that both mining companies and equipment manufacturers stand to benefit: mines achieve higher production and improved safety by removing operators from the pit, while OEMs gain insight into which automated functions deliver the most value for future machine designs. It provides a quantitative case for ongoing investment in automation technology.

UGV navigation underground (review): complementing the above, a comprehensive review by Abdukodirov and Benndorf (Contribution 7) surveyed the state of path-planning algorithms for unmanned ground vehicles (UGVs) in underground mines. Reliable navigation is essential for autonomous trucks, loaders, or inspection robots to operate in the dark, GPS-denied, and cramped conditions of an underground mine. The review outlines the key requirements for effective underground path planning (robust sensors, mapping systems like SLAM, and considerations for the Robot Operating System) and examines a range of algorithmic approaches. Both global path planners (graph-based algorithms such as A* and Dijkstra, and sampling-based methods such as RRT) and local planners (real-time obstacle avoidance using potential fields, ant colony algorithms, or reinforcement learning) are discussed, along with their respective strengths and typical use cases. Notably, the review identifies a gap between simulation and reality: many algorithms have been tested only in research simulations and face challenges such as computational load and sensor noise when applied in actual mines. It concludes with a call for full-scale field trials in real mines and possibly hybrid approaches that combine the best of different algorithms. The insights from this paper guide practitioners in identifying the most promising navigation techniques and underscore the need to move beyond theoretical research to truly enable autonomous underground vehicles.

4. Innovations in Monitoring and Sensing Technologies

Accurate monitoring of mining processes and geotechnical conditions is crucial for efficiency and safety. This Special Issue covers studies which introduce novel approaches for sensing and measurement, which improve our ability to gather data in challenging mining environments:

Three-dimensional fragmentation measurement via drone: one technical contribution by Toriya et al. (Contribution 8) introduced an improved method for measuring blast fragmentation size distribution using drone-based photogrammetry enhanced with GNSS (satellite positioning). Fragmentation (post-blasting rock size) is traditionally assessed by manual sampling or photographs with reference scales, which can be tedious or inaccurate. The researchers developed a system to automatically generate a scaled 3D model of a post-blast rock pile using only the positional data from the drone’s GNSS, without any ground control points (GCPs) or manual scale objects. Field trials at an operating mine showed that this GNSS-aided approach successfully produced accurately scaled rock pile models, enabling reliable particle-size analysis. In surface mines where GPS signals are available, this technique eliminates the need for surveyors to access the muck pile (enhancing safety) and accelerates the survey process. The result is near-real-time feedback on blasting outcomes, which assists engineers in improving blast designs for optimal fragmentation (thereby improving downstream efficiencies). This work exemplifies how combining drone imagery with precise positioning data can automate a task that once required considerable human effort.

Slope stability monitoring (review): ensuring the stability of large open-pit slopes is a vital safety concern, and this Special Issue includes a systematic review of slope monitoring methods and technologies conducted by Le Roux et al. (Contribution 9). The review covers both traditional in situ instruments (extensometers, inclinometers, piezometers, etc.) and modern remote sensing tools (radar interferometry, LiDAR, satellite and drone photogrammetry), as well as emerging digital technologies like IoT sensor networks and AI-based data analysis. By comparing these approaches across attributes such as accuracy, spatial/temporal coverage, complexity, and cost, the review finds that no single monitoring method is universally optimal. Each technology involves trade-offs: for example, radar and satellite methods provide broad coverage but may be limited by weather or update frequency, whereas ground sensors give precise continuous data but only at discrete points. To overcome these limitations, the review suggests future systems must integrate multiple sensing methods, potentially combining remote and in situ data through IoT platforms and using AI to fuse and interpret the information in real time. Such an integrated multi-sensor approach would enhance early-warning capabilities for slope failures. The review serves as a consolidated knowledge base on slope monitoring, guiding mines in designing robust monitoring frameworks and highlighting research directions to develop cost-effective, scalable solutions that improve mine safety.

5. Alternative and Integrated Mining Approaches

Innovative techniques and integrated planning frameworks are also discussed, reflecting the drive to make mining more sustainable and versatile:

Combined conventional and in situ mining: Krassakis et al. (Contribution 10) examined how conventional mining might be combined with an alternative extraction method, specifically, Underground Coal Gasification (UCG), which converts coal to gas in situ without physical mining. The paper detailed the construction of a 3D geological model of the Kardia lignite deposit and used GIS-based analysis to identify zones that met the criteria for UCG (sufficient seam thickness, appropriate depth, etc.). By mapping these zones, the researchers showed where UCG could potentially be implemented alongside or after the open-pit operations. This integrated geospatial approach helps pinpoint suitable areas for unconventional methods and supports planning a combined coal-to-liquid supply chain within a complex deposit. The broader significance is that such modelling may extend mine life: uneconomical resources that may not be economically viable by traditional methods might economic via techniques such as UCG. Moreover, the study’s workflow—building a detailed subsurface model and overlaying alternative method criteria—can save time and cost in evaluating novel mining options, contributing to more sustainable resource utilisation. It underscores the value of advanced geological modelling in decision-making, especially when considering the adoption of emerging mining methods.

6. Broader Perspectives and Future Directions

Two contributions provide a broader perspective on how technology is reshaping mining through extensive literature reviews:

Real-time mining (RTM) review: a comprehensive review by Anvari and Benndorf (Contribution 11) on real-time mining (RTM) examined developments over the last decade that enable mines to adapt operations dynamically based on continuous data. RTM is presented as a response to pressing industry challenges, including lower ore grades, deeper deposits, and higher energy costs, and aims to integrate advanced sensing and analytics to enable immediate decision-making. The review highlights key components of RTM: real-time grade monitoring, dynamic mine planning and scheduling that adjust to new information, and continuous optimisation algorithms that refine production. The paper discusses the role of technologies such as high-resolution online sensors, data assimilation techniques (e.g., Kalman filters), and AI in updating mine models on the fly. Importantly, the authors document how RTM can enhance sustainability and profitability simultaneously by extracting resources more efficiently by reducing waste and energy use, and enabling quicker responses to operational issues, thus minimising downtime. The review proposed a comprehensive framework for adaptive, data-driven mining operations that leverage continuous feedback to support sustainable development and better decision-making. RTM is positioned as a cornerstone of future mining, drawing parallels to the Industry 4.0 revolution in manufacturing and suggests future mines will operate as interconnected, intelligent systems.

Emerging technologies in mining (bibliometric study): another article (Emere et al., Contribution 12) offers a bibliometric analysis of emerging technologies (ETs) in the mining sector, quantitatively tracking research trends from 1986 to 2021. By examining 135 scholarly publications on topics like AI, digital twins, Internet of Things (IoT), and blockchain in mining, the study visualises how interest in these technologies has grown and which areas are most influential. The analysis confirms a significant surge in adoption and research of digital and automation technologies, reflecting the mining industry’s drive to improve safety, efficiency, and sustainability through innovation. Key clusters identified include automation and AI in mining operations, smart mining initiatives, and the use of blockchain for supply chain transparency. The study provides stakeholders with a “state-of-the-art” map of technology integration in mining, and it suggests that the sector has reached a tipping point where such technologies are becoming mainstream. The bibliometric review thus not only charts past and present trends but also points to cross-industry learning as a strategy to accelerate mining’s technological transformation.

7. Conclusions

The Special Issue papers paint a picture of an evolving mining industry, one that is increasingly data-driven, automated, and technologically innovative. From the detailed case studies to the broad reviews, a common thread is the pursuit of safer and more efficient mining through advanced modelling and technology deployment. Several recurring insights emerge:

• Integration is key: no single technology or solution can address all mining challenges. Many contributions highlight the need for integrated approaches, whether combining multiple monitoring techniques for slope stability or integrating human decision-makers into automated systems to achieve optimal results. The future mine will likely be a synergistic blend of AI algorithms, robotics, sensors, and human expertise working in concert.
• From reactive to proactive: tools like predictive ML models and real-time analytics are shifting mining from a reactive mode (responding to problems after they occur) to a proactive, anticipatory mode. For instance, anticipating equipment wear from material properties, or adjusting plans on the fly through real-time mining, can prevent issues and seize opportunities earlier. This predictive capacity will be crucial as mines operate in increasingly complex and challenging conditions.
• Safety and sustainability: underlying many of these works is the twin goal of making mining safer for people and less impactful on the environment. Automation and remote operation directly remove people from harm’s way, whether by autonomous vehicles or zero-entry concepts. At the same time, optimised processes and better monitoring mean less waste and lower emissions (e.g., efficient blasting reduces vibration and noise, and precise extraction reduces energy use). The Special Issue shows that technology can enable the mining sector to meet production demands while also adhering to stricter safety and sustainability requirements.
• Bridging the gap to implementation: a note of pragmatism is visible as well. Several authors point out that despite promising results, challenges remain in moving from research to real-world deployment. Whether it is the need for more on-site trials of autonomous navigation algorithms or ensuring machine learning models remain robust outside their training data, there is work to do to translate these advances into everyday mining practice. Change management and workforce training will also play a role as the mining workforce adapts to new roles (such as managing automated systems or interpreting complex data outputs).

In summary, this Special Issue highlights the potential of combining modern technology with mining engineering. The contributions demonstrate advances, higher prediction accuracy, productivity gains, improved monitoring, and comprehensive frameworks for future operations. Importantly, they suggest the future of mining will be defined by connectivity and collaboration: connecting disparate data sources in real time, connecting people with intelligent machines, and connecting cross-disciplinary innovative ideas to solve mining’s challenges. The insights and directions outlined will assist the mining industry in moving towards more sustainable, safer, productive, and cost-effective operations. However, further investigations are required, particularly within alternative/future mining methods, communication systems, mine electrification, and advanced sensor technologies. The authors encourage researchers to submit their latest research findings in these areas to the second edition of this Special Issue [4].