Mobile, Handheld & SLAM Scanning

To understand mobile scanning you have to appreciate a chicken-and-egg problem. A stationary terrestrial scanner knows exactly where it is — it is bolted to one spot — so every point it measures can be referenced to that fixed origin with confidence. A moving scanner has no such luxury. Each instant it is somewhere new, and it does not have a GPS-accurate position indoors; yet to assemble its measurements into one map, it must know where it was when it took each one. But working out where it is depends on recognising the surroundings it has already mapped. So it needs the map to find its position, and its position to build the map. This is the SLAM problem: simultaneous localisation and mapping.

The resolution is to do both at once, continuously, each refining the other. As the scanner moves, it watches how the surrounding geometry shifts between successive instants and infers how it must have moved to produce that change — estimating its own trajectory, a running chain of positions and orientations (poses). At the same time it places its new measurements into the growing map using that estimated trajectory. Good SLAM systems fuse several sensors to do this well: the LiDAR itself, an inertial measurement unit (accelerometers and gyroscopes sensing motion, like the ones in your phone), sometimes cameras, all blended so the estimate is more robust than any one sensor alone. The phone in your pocket does a simpler version of the same thing to track itself in augmented reality.

The profound consequence is that the map is built relative to the scanner's own journey, not to a fixed external point. This is what frees you to keep moving — and, as the next section explains, it is also the root of the method's characteristic error. SLAM is a triumph of estimation: it makes continuous, walk-and-scan capture possible, and it powers not just mobile scanners but self-driving cars and robot vacuum cleaners. But it is estimation, and the honest habit of this course applies with special force here — the position of every point rests on a *computed guess* of where the scanner was, and those guesses, however clever, carry and accumulate uncertainty.

Because SLAM builds the map by chaining together estimates of how the scanner moved from one instant to the next, small errors in each step do not simply average out — they accumulate. A tiny misjudgement of motion early on shifts everything that comes after, the next small error adds to it, and over a long walk the estimated trajectory can gradually wander away from the truth. This accumulating error is called drift, and it is the defining limitation of mobile scanning. Walk a long corridor and back with a naive system and the two ends of your path, which should meet, may not quite line up; the map has bent slightly over its length.

The principal cure is elegant and worth knowing: loop closure. When the scanner recognises that it has returned to a place it mapped earlier — closing a loop in its path — the software gains a powerful new constraint: these two points in the trajectory must actually be the same place. It can then redistribute the accumulated error back around the whole loop, snapping the drifted map back into consistency. This is why good mobile-scanning practice involves planning your route to close loops deliberately: returning to the start, crossing earlier paths, and overlapping between floors. A walk with well-closed loops is far more accurate than an out-and-back that never revisits anything.

Even with loop closure, the honest comparison stands: mobile and SLAM scanning is generally looser in accuracy than a terrestrial scanner. A stationary TLS references every point to one fixed, stable origin with survey-grade encoders; a mobile system references everything to a continuously estimated, drifting trajectory. So where a TLS survey might be trusted to a fine tolerance confirmed by its spec, a mobile scan is typically a coarser, though often perfectly adequate, record. As always the actual numbers are verified instrument specifications under stated conditions, not rules of thumb — but the *direction* of the trade-off is reliable and central: you gain enormous speed and coverage and give up some accuracy and crispness. For many jobs that is an excellent bargain; for jobs needing the tightest geometry, or anything binding, it is not, and you either use TLS or control the mobile work against surveyed points and a licensed surveyor.

Mobile scanning comes in a family of form factors, all sharing the SLAM engine but packaged for different jobs. Handheld scanners are the most familiar: a device about the size of a small camera or a torch that you carry and wave as you walk through a space, scanning continuously. They are quick to deploy, nimble in tight and cluttered interiors, and ideal for capturing rooms, stairs and awkward corners that a tripod would labour over. Backpack systems mount the scanner (often with multiple sensors) on a frame you wear, leaving your hands free and letting you capture large areas — multiple floors, long building circuits, campuses — at walking pace, which is transformative for big interiors. Trolley or cart-mounted systems roll the scanner along floors and corridors, excellent for large, relatively flat environments such as warehouses, offices, hospitals, stations and tunnels, where smooth wheeled motion also gives the SLAM a steady ride.

Beyond these, the same principle scales up to vehicle-mounted mobile mapping systems — scanners on cars or rail vehicles, usually fused with satellite positioning (GNSS) outdoors — that capture streets, roads and rail corridors at driving speed, a cornerstone of large-scale infrastructure and city mapping. The common thread across the whole family is capture on the move: rather than a sequence of discrete stations, you record a continuous stream of points along a path, and SLAM (indoors) or SLAM fused with GNSS (outdoors) knits it into a map.

Choosing within the family follows the job. Handheld for agility and tight, detailed spaces; backpack for covering large or multi-level interiors fast and hands-free; trolley for big flat runs; vehicle-mounted for streets and corridors. In every case you accept the SLAM accuracy bargain in exchange for speed, and you plan your route to close loops and maintain steady, overlapping motion. A growing and practical pattern is hybrid capture: use a fast mobile or handheld scan to cover the bulk of a large space quickly, then set up a terrestrial scanner at the few locations that demand the tightest accuracy, and combine the two. That plays each tool to its strength — broad coverage from the mobile device, anchor accuracy from the TLS — and is increasingly how large real-world jobs are actually run.

Mobile and SLAM scanning earns its place wherever speed and coverage matter more than the last increment of accuracy, and there are many such jobs. Large interiors are the classic win: a big office floor, a shopping centre, a hospital wing, a factory hall — spaces where a terrestrial scanner would need dozens of slow stations but a backpack or trolley can capture at walking pace in a fraction of the time. Corridors and repetitive circulation — long hospital or hotel corridors, basements, service routes, multi-storey stairwells — suit mobile scanning especially well, because they are tedious and station-heavy for a TLS but a natural continuous walk for a mobile system. Fast walk-throughs for coordination, space planning, facilities documentation, retail or hospitality rollouts, and periodic re-capture to track change all benefit from the sheer speed of getting a whole building as a usable cloud in one visit.

Mobile scanning is also valuable where access is awkward or time on site is tight — a live, occupied building you cannot shut down for a long terrestrial survey, or a site where you simply cannot linger. And it pairs naturally with the hybrid approach: mobile for the bulk, terrestrial for the few accuracy-critical areas.

The honest counter-list matters just as much. Stay with TLS (or control the mobile work carefully) when you need the tightest accuracy — fine heritage detail, precise as-builts for close-tolerance fabrication, deformation or structural checks — because the drifting trajectory behind a mobile scan cannot match a stationary survey-grade instrument. Small, detail-rich spaces where you are not really moving much lose the speed advantage and may be better served by a TLS or even a careful handheld with tight loops. And, as ever, anything legally or structurally binding, georeferenced or survey-grade is the province of a licensed surveyor working to verified methods, whatever instrument is used underneath. The professional skill is to read the job honestly: if the prize is covering a lot of space quickly to a reasonable accuracy, mobile scanning is often the smartest tool in reality capture; if the prize is guaranteed tight geometry, it is not the tool to trust alone.