Key Takeaways

  • Automated ore sampling is upgrading shovel-face, sample-prep and blasthole/RC sampling with sensor packages that reduce manual handling and speed up grade turnaround.
  • Motion Metrics ShovelMetrics and MineSense’s shovel sensor line use AI and computer vision at the dig face in real time.
  • Scott Automation’s Rocklabs line (AMS Prep, RoboPrep Elite) automates crushing, splitting and pulverizing, so cross-belt or lab autosamplers can feed inline XRF without manual prep.
  • IMDEX BLASTDOG instruments blasthole/RC sampling, capturing sensor-while-drilling data as the hole is drilled.
  • Blasthole/RC sampling AI is the least mature of the three (Emerging). Most of the intelligence currently sits in the downstream XRF/assay step rather than the sampling mechanism itself.

TL;DR

Automated ore sampling puts a sensor as close to the point of extraction as possible: in the shovel bucket, on the prep line or down the blasthole. AI models then read that signal continuously instead of waiting for a bagged sample to reach the lab. As a result, the grade-control decision loop shrinks from days to minutes.

What Is Automated Ore Sampling?

Ore sampling means taking small, representative portions of broken rock so a lab can measure grade. Traditionally, a shovel operator digs, someone bags a sample, and results come back hours or days later. Sampling sits at the front of every grade-control and geometallurgy pipeline, so it is often the slowest link.

Automated ore sampling replaces those manual steps with machines and sensors. Robots crush and split the material, and sensors read the rock directly where possible. The goal is the same as manual sampling: a representative measurement. However, automation delivers it faster, more consistently and with fewer people exposed to the pit floor.

How Do I Automate Ore Sampling With AI?

The current generation of tools attacks each stage of the chain separately. Start with the stage that causes the most delay at your site.

Sensing at the dig face

Shovel-mounted systems such as Motion Metrics’ ShovelMetrics (now part of Weir) use 3D cameras and computer vision. They continually estimate fragmentation and flag oversized material directly from the bucket, without stopping the dig cycle. MineSense’s shovel sensor products work at the same point. This is a genuinely real-time AI application, because the model runs on every pass rather than on a periodic sample. Grade sensing in the bucket is covered in more depth in real-time ore grade sensing at the shovel.

Automated ore sampling in the prep line

For sample preparation, Scott Automation’s Rocklabs line includes the AMS Prep Line and RoboPrep Elite. These systems robotically crush, split, dose and pulverize samples with RFID tracking. They feed cross-belt autosamplers with inline XRF analyzers. The AI contribution here is less about the mechanics. Instead, it comes from XRF analyzers that use dynamic, matrix-aware calibration to hold accuracy across variable ore types, as described in our page on grade control sensors.

Instrumenting blasthole and RC sampling

Blasthole and RC sampling has a more specific tool: IMDEX BLASTDOG. It is an eight-arm calliper and multi-sensor probe that rides down the blasthole on a semi-autonomous tracked platform. It captures hole diameter, voids and structure as sensor-while-drilling data, rather than relying purely on the recovered sample. However, this is the least mature stage for AI. Standard RC rig cone splitters are still the default, and BLASTDOG-style instrumentation is layered on top rather than replacing them. Read more in our BLASTDOG blasthole logging page.

What to Expect From Automated Ore Sampling

Across all three stages, the practical move is the same. Wherever you can put a sensor closer to the point of extraction, you shorten the grade-control decision loop.

Sensor readings still need to be trusted, though. Therefore, plan to reconcile them against lab assays for months before using them alone. Sampling theory still applies too: a fast reading of a poorly split sample is still a poor reading. Keep duplicates, blanks and standards in the stream, just as you would for manual sampling.

FAQs

What is automated ore sampling?

Automated ore sampling uses robots and sensors to collect, prepare and measure ore samples with minimal manual handling. It covers shovel-mounted sensors, robotic sample-prep lines and instrumented blasthole probes. The aim is faster, more consistent grade data.

Does automated ore sampling remove the need for a lab?

No. Most sites still send check samples to a certified lab. Sensor and inline XRF results are fast, but they need regular reconciliation against lab assays to stay accurate. The lab becomes a verification step rather than the bottleneck.

Where does AI add the most value in ore sampling?

AI adds the most value at the dig face and in the analyzer. Computer vision reads fragmentation from every shovel pass, and matrix-aware XRF calibration holds accuracy across changing ore. The mechanical prep robots themselves mostly follow fixed routines.

Is blasthole sampling automation mature?

Not yet. Instrumented probes like BLASTDOG are commercially available, but the AI layer on blasthole and RC sampling is still Emerging. Cone splitters on standard rigs remain the default method at most mines.

Try It With Geocluster

Sensor-based sampling generates a lot of streaming, semi-structured data. That data needs to be reconciled against lab assay results before anyone trusts it. This is exactly the research and validation work Geocluster is designed for. If you are evaluating or deploying sensor-based sampling, Geocluster can help you build the reconciliation pipeline behind it.