All articles
Energy / Research insight

Substation Digital Twins: Build the Map Before Automating the Inspection

AutoMap3 min read

A substation inspection becomes more useful when the team can return to the same asset, understand its surroundings and compare observations over time. SLAM can provide the spatial foundation, but a navigable point cloud is only one part of a dependable inspection workflow.

Illustrative point cloud of an industrial plant
Illustrative imagery from the AutoMap library; not an image from the cited studies.

What to take into your next project

  • Separate mapping, navigation and condition assessment.
  • Test transitions and uneven terrain, not just clear access roads.
  • Connect observations to persistent asset identifiers.

Start with the inspection question

Consider a maintenance team revisiting a transformer bay after an abnormal thermal reading. They need to identify the equipment, reproduce the observation and understand what has changed. A sensible mapping brief therefore begins with the inspection records the team needs to produce. Coverage, asset naming and repeatable observation locations deserve attention alongside the appearance of the 3D model.

What the substation research shows

A 2023 Energy Reports study combines LiDAR and inertial measurements through a tightly coupled positioning system. Its processing includes correction for motion during scanning and optimisation over a moving window of observations. The lesson for a purchaser is to assess the complete sensing and processing chain: movement and timing affect the map before anyone opens it in a viewer.[1]

Research published in Energies in 2025 adds visual information to LiDAR estimation and generates a colour point cloud. The work addresses environments where geometry alone provides weak or ambiguous positioning clues. Colour also gives a human reviewer useful context, although a recognisable item in an image still needs a reliable connection to the asset register.[2]

A separate ground-vehicle study considers rough substation terrain, localisation within an existing map and extensions into previously unmapped areas. It also processes the ground to identify traversable regions for inspection. This makes a useful distinction: a surface can be visible in a scan without being a suitable route for a robot.[3]

Turn the findings into an acceptance plan

Our practical recommendation is to commission the map and the inspection workflow together, while checking their outputs separately. Select a pilot area containing ordinary equipment bays, repeated structures and the transitions the team actually uses. Include uneven surfaces and locations with restricted views. Keep the operator-approved access arrangements in place throughout the trial.

  • Check map alignment against independently surveyed locations.
  • Ask the inspection team to locate named assets without help from the scan operator.
  • Repeat selected observations and record whether the same view can be reproduced.
  • Record obscured areas and uncertain matches instead of treating them as complete.

Make the handover useful

Package the accepted map with its capture date, reference system, coverage notes and asset links. Keep thermal images, photographs and other condition measurements identifiable as separate observations. A later reviewer should be able to tell whether a change reflects the equipment, the viewing angle or a map update. This turns a promising visualisation into a working reference for the maintenance team.

For an AutoMap project, bring an example inspection report to the initial discussion. It helps define what the spatial data must support and which additional sensors or integrations the workflow will require.

Sources & further reading

This original article draws on the research below. Practical recommendations are editorial synthesis; the cited studies are not performance claims for AutoMap products. Sources checked on 10 September 2026.

  1. 01
  2. 02
  3. 03

Map your own challenging environment.

Tell us about your site and the data you need. We’ll discuss a capture setup and a practical way to evaluate the results.

Talk to us