SLAM vs terrestrial laser scanning: what 8 practitioners say

At a glance

How much accuracy does the job really need?

Almost every guest started from the same question: what will the data be used for? Michael Schoelkopf of NavVis points out that floor plans and LOD 200 work don’t need millimetres, and that high-end scanners can be overkill for low-margin jobs. Ben Williams of Exyn Technologies says that in Exyn customers’ projects most of the data doesn’t need one-millimetre accuracy.

The same message comes from very different corners: Zlata Karpushina of FJ Dynamics says the first step is understanding what accuracy the client needs, Rami Tamimi says asset inventory doesn’t need the highest accuracy, and Dylan Gorman makes the same point about earthworks. Legal and design deliverables are different – there, Rami Tamimi says, different accuracy standards and a licensed surveyor still apply.

Drift: SLAM’s real weakness – and TLS isn’t immune

Rami Tamimi describes SLAM error as compounding: the longer the capture, the more error accumulates, and the algorithms are less stable than a static setup. Dylan Gorman calls drift one of SLAM’s biggest problems and attributes it mostly to the algorithms rather than the sensor.

But the guests are clear that terrestrial scanning has its own version of the problem. Rami Tamimi says total stations and terrestrial scanners also compound error setup after setup, and Kelly Watt recalls a TLS project where areas ended up 14 feet off because survey control wasn’t used effectively.

Speed: where SLAM wins

Speed is SLAM’s core argument. Zlata Karpushina sums up its value as speed, mobility and flexibility, and describes a capture as a single walk of several minutes. Michael Schoelkopf gives the example of apartments of 50–100 m² scanned in about five minutes each. Even Rami Tamimi, the strongest TLS advocate among the guests, agrees that in many applications a terrestrial setup is simply the slower workflow.

Michael Schoelkopf adds a business caveat: time saved in the field only turns into saved cost if you have enough work to fill that time. And Povilas Sindriūnas of Undet notes that once capture gets faster, the bottleneck moves to the office – registration and modelling.

When SLAM is the wrong tool

SLAM needs geometry to know where it is. Michael Schoelkopf says any SLAM device fails where there is none – open outdoor areas or tunnels – and that Michael Schoelkopf would not use a compact handheld for a 10-metre ceiling or LOD 300 detail. Zlata Karpushina lists long corridors, repetitive spaces, glass, reflective surfaces, open areas and poor capture routes as the conditions where drift and misalignment still happen.

For millimetre-level, controlled surveys, Zlata Karpushina – speaking for a SLAM vendor – would still reach for a terrestrial scanner – and Rami Tamimi says survey-grade underground work still means a total station traverse and TLS on targets.

Control points, GNSS and combining both

Nobody recommends uncontrolled SLAM on large sites. Michael Schoelkopf says loop closures are always recommended and that big sites like factories still use a control-point network. Mehdi Boroumand works with at least five ground control points without RTK – or RTK/PPK on the scanner instead. Rami Tamimi’s recipe for a tunnel that isn’t survey-grade: RTK positions at the start and end on the surface, SLAM underground, then adjust.

The most practical answer is often both. Ben Williams describes customers building the base map with SLAM and using TLS only where precision is critical, and Kelly Watt points out that control points work indoors too.

Training and the “one-button” myth

This is where the guests disagree most. Michael Schoelkopf believes anyone can scan a small apartment with a handheld SLAM device, while experience still improves results. Rami Tamimi sees SLAM as harder for beginners – more sensors, more variables – and Kelly Watt would have people learn terrestrial scanning first.

Interestingly, the strongest warning comes from a SLAM vendor: Zlata Karpushina calls the one-button promise one of the most dangerous in technology because it breaks expectation management, and says there is a learning curve that honest marketing should admit.

Will SLAM replace TLS?

Opinions split. Rami Tamimi expects terrestrial scanning to keep dominating survey work. Povilas Sindriūnas sees no viable alternative to SLAM for as-built documentation, Kelly Watt expects a lot more SLAM, and Ben Williams believes the specific role of carrying a tripod scanner around a site will go away as autonomous systems take over.

On what will decide the market, the SLAM side agrees: Zlata Karpushina and Mehdi Boroumand say the software matters more than the sensor. Michael Schoelkopf partly disagrees – in Michael Schoelkopf’s view the LiDAR itself sets the limits, and a new generation of sensors would be the real step change.

Questions

Is SLAM accurate enough for surveying?

For reference data, floor plans and LOD 200 modelling, the guests say SLAM is usually enough. For survey-grade, legal or millimetre-level work, Rami Tamimi and Zlata Karpushina (FJ Dynamics) would choose a terrestrial scanner with proper control.

How do you reduce SLAM drift?

Guests recommend loop closures, control points (Mehdi Boroumand uses at least five GCPs), GNSS/RTK fixes at the start and end of a capture, and avoiding featureless spaces, long corridors and glass.

Should a beginner start with SLAM or a terrestrial scanner?

Kelly Watt and Rami Tamimi recommend learning terrestrial scanning first; Michael Schoelkopf (NavVis) says small SLAM jobs are easy to start, but experience still improves results.

Can SLAM and TLS be combined on one project?

Yes – Ben Williams (Exyn) describes customers building the base map with SLAM and using TLS where precision is critical, and NavVis IVION accepts TLS data next to SLAM data, per Michael Schoelkopf.