handovers, supervisor routines, and management reviews, with digital KPIs treated as core operating metrics rather than supplementary reports. 7. OUTLOOK: THE AUTONOMOUS, INTELLIGENT UNDERGROUND MINE The deployment experience reviewed in this paper suggests that the autonomous underground mine will not emerge from a single advanced technology, but from the gradual integration of connectivity, telemetry, analytics, and operating discipline. The most important learning is that autonomy depends first on visibility. Underground mines cannot optimize what they cannot consistently measure, and even advanced AI models have limited value if equipment data is intermittent, fragmented, or disconnected from production workflows. For this reason, the first practical step toward autonomy is not full automation, but reliable data capture across the mobile fleet, haulage routes, maintenance areas, and critical production locations. A second learning is that digital systems create value only when they are embedded into daily decision-making. Fuel analytics, maintenance analytics, and production dashboards are most effective when dispatchers, supervisors, maintenance planners, and site managers use them during the shift, not only after the fact. In practice, this means that the path toward intelligent mining begins with relatively practical use cases: identifying non-productive fuel consumption, reducing queues, structuring maintenance logs, detecting early signs of component failure, improving tire management, and comparing cycle performance across routes and crews. These use cases build trust in the data and create the operating habits needed for more advanced AI-enabled decisionmaking. A third learning is that optimization must be system-wide. Underground mines are constrained environments where changes in one area quickly affect another. A routing change may reduce fuel consumption but increase congestion; a maintenance intervention may improve availability but disrupt short-term production; a higher haulage rate may expose bottlenecks at loading, dumping, or crushing points. The value of an Operating System of Intelligence is therefore not simply that it produces more dashboards, but that it connects fuel, maintenance, equipment health, location, and production data into one operating context. This allows mines to understand trade-offs and manage the underground value chain as a coordinated system. These lessons indicate that the future autonomous mine will likely develop in stages. The first stage is connected visibility; the second is data-driven intervention; the third is predictive and prescriptive control; and only then does broader autonomy become realistic. In this progression, every litre of fuel, every tonne moved, every component replaced, and every delay in the production cycle becomes part of a continuous improvement loop. The practical implication for operators is clear: mines that invest early in reliable connectivity, structured data, and workflowbased analytics will be better positioned to adopt autonomous and self-optimizing systems than those that treat digital tools as isolated pilots. REFERENCES
RkJQdWJsaXNoZXIy MTM0Mzk2