Verify what was built

Measure the outcome, not the spectacle

A robot’s repeatability is not the same as a building’s dimensional accuracy. Repeatability describes how consistently a system returns to a condition under defined circumstances. Installed accuracy also depends on calibration, workholding, material behaviour, reference geometry and assembly. Ask a supplier to identify exactly what its quoted metric describes and how that metric relates to the completed component.

Even excellent dimensional control does not prove fire resistance, water resistance, acoustic separation or thermal performance. Each characteristic needs an appropriate design basis and verification route. The inspection plan should distinguish measurements that confirm workmanship from tests that assess an assembly’s performance. This prevents a detailed point cloud from being mistaken for a comprehensive certificate of quality.

Agree a reference that can survive the project

Survey data needs a stable coordinate system and a clear relationship to the approved design. Preserve control points where possible, record transformations and identify who accepts the reference network. When a scan is compared with a model, the alignment method matters: an apparently good fit can obscure a local deviation if the registration process redistributes error.

Point clouds also have gaps, noise and limits of access. A scanner cannot inspect the concealed side of a connection merely because its visible face has been captured. Define coverage and uncertainty, then supplement the scan with targeted checks. For an exposed timber junction, that might include direct measurement of a bearing surface before the neighbouring member is installed.

Use mock-ups to resolve judgment

Some premium residential requirements are difficult to reduce to a single dimension. Raking light reveals slight changes in surface alignment; adjacent materials change the perceived colour of a finish; a handle’s operation depends on tactile resistance as well as geometry. Agree a representative mock-up under relevant lighting and retain a record of the accepted condition.

A mock-up should include the interfaces that can fail. A window sample is more useful when it includes the wall build-up, sill, drainage and interior return. Where practical, test the assembly before committing to the full package. Record which aspects the sample approves: visual appearance, workmanship and performance are related, but approval of one should not silently approve all three.

Test building performance separately

ISO 9972 describes the fan-pressurisation method for determining air permeability of buildings. It provides a test method, not a universal target for every home. The project’s required performance must come from the applicable brief, regulations and design strategy. A test is most useful when planned early enough that leakage can still be found and repaired.

The Passivhaus Institute publishes criteria and certification resources for its voluntary building standard. Those criteria can be relevant when a project specifically adopts that route; robotic fabrication does not confer Passivhaus status. The thermal envelope, ventilation system, junction detailing and verification process must work together, regardless of how the components were made.

For acoustics, the ISO 16283 series concerns field measurement of sound insulation in buildings and building elements. Again, the chosen method and applicable target need competent interpretation. A tightly machined panel joint does not by itself prevent flanking transmission through a floor, service route or continuous structural member.

Keep environmental claims within their boundaries

EN 15978 provides a framework for assessing environmental performance at building level, while EN 15804 establishes core rules for construction-product environmental declarations. These are relevant when testing claims that robotic fabrication reduces impact. A reduction in offcuts is one contribution, not a whole-life assessment.

Use comparable functional requirements, study periods and system boundaries. Include materials added to make the process work, transport to specialist facilities and plausible replacement scenarios. The RICS Whole Life Carbon Assessment for the Built Environment offers a structured methodology for assessment. None of these frameworks makes an environmental conclusion automatic; the result depends on the data and assumptions used.

Hand over a maintainable home

The final ledger should connect installed components to inspection results, approved deviations and maintenance instructions. Retain useful production identifiers without overwhelming the homeowner with machine files they cannot interpret. Identify specialist coatings, replaceable elements and any restrictions on drilling or alteration.

BSRIA’s Soft Landings approach emphasises the transition from construction into use. That is particularly relevant when a home contains unfamiliar assemblies or tightly coordinated services. Arrange commissioning, understandable user guidance and a route for reviewing early operational issues. Living quality is tested over time, not exhausted at practical completion.

The strongest verification plan therefore ends with a feedback loop. If an unexpected movement, leak or maintenance difficulty appears, the team should be able to trace the affected detail and assess comparable locations. Robotics can support that traceability, but only if the information was deliberately retained and assigned to someone responsible for using it.