Has the Ground Changed? Reading Repeat SLAM Surveys in Underground Mines

Begin with the change you need to detect

A large missing piece of shotcrete and a small gradual displacement are different measurement problems. A workflow capable of identifying the first does not automatically establish the second. Before capture, describe the area of interest, expected change, monitoring interval and the decision the result will inform. That gives the team a basis for judging whether the data is suitable.

Quality exists at more than one scale

An ARMA paper on mine-wide mobile LiDAR workflows separates target-level quality from site-level quality. It examines accuracy, density, coverage and drift rather than relying on one number. It also explains that aligning scan sections can improve local agreement while potentially obscuring broader changes. Registration is therefore part of the measurement argument, not merely a cosmetic preparation step.[1]

A subsequent Mining, Metallurgy & Exploration study evaluates SLAM data for underground geotechnical monitoring and demonstrates application-specific processing for change detection. Its results support the potential of mobile scanning, while showing why suitability depends on the collection and processing workflow. They do not establish that every mobile scan is fit for every deformation task.[2]

An evaluation of LiDAR-based SLAM algorithms in a SubT environment compares both estimated trajectories and reconstructed tunnel maps. That dual assessment is useful for monitoring projects: an apparently plausible route does not, by itself, establish that the surfaces being compared have been reconstructed consistently.[3]

Treat comparison as an investigation

Our practical recommendation is to preserve enough information to explain a difference. Record the reference frame, capture route, processing version and coverage for each survey. Use suitable independent checks and clearly defined stable references. Be cautious about allowing the surface under investigation to determine all of its own alignment, because that can reduce the apparent change being measured.

• Check whether both surveys actually observed the same surface.
• Inspect vegetation, vehicles, stored material and other temporary objects separately.
• Keep raw or minimally processed data alongside derived comparison outputs.
• Refer significant findings for the project’s established geotechnical review.

Present evidence a reviewer can follow

Imagine that one side of a pillar appears to have moved while the nearby floor also shifts by a similar amount. Before attributing this pattern to ground behaviour, the reviewer should examine alignment and observation conditions. A second view or a targeted independent measurement may clarify the cause more effectively than adjusting the colour scale.

The final deliverable should connect each flagged area to the underlying observations and the limits of the comparison. SLAM can extend the spatial record available to a mine team, while the interpretation of ground behaviour remains a qualified engineering task. For an AutoMap trial, agree those review requirements before collecting the first baseline.