SLAM mapping explained: handheld lidar scanners and mobile mapping
Key takeaways
- SLAM mapping estimates the scanner’s path and the map together, so no GNSS signal or tripod setups are needed.
- Lidar SLAM is the standard for survey work; visual SLAM is cheaper but more sensitive to light and texture.
- Drift is the main accuracy risk, and closed loops plus control points are the main defences.
- Handheld SLAM scanners are typically several times faster than terrestrial scanning, with lower but often sufficient accuracy.
- Plan the walking route as carefully as you would plan tripod stations.
What is SLAM mapping?
SLAM mapping stands for simultaneous localisation and mapping: a moving sensor builds a map of an unknown environment while estimating its own position in that map at the same time. It solves a chicken-and-egg problem, because accurate mapping needs a known position, and positioning needs a map.
SLAM technology began in robotics, where robots need to navigate indoors without GNSS. In surveying, it powers handheld and backpack laser scanners that capture buildings, tunnels, mines and forests as the operator walks. The output is a point cloud similar to that from a tripod scanner, but captured continuously.
SLAM is one form of mobile mapping, the wider term for capturing spatial data from a moving platform. Our article on how lidar scanners work covers the sensor side of the technology.
How does a SLAM algorithm work?
A SLAM algorithm repeatedly predicts where the scanner has moved, checks that prediction against what the sensor sees, and updates both the path and the map.
The process has two main parts. The front end handles short-term tracking: an inertial measurement unit (IMU) predicts motion over fractions of a second, and scan matching aligns each new lidar sweep to the recent map, refining the pose (position and orientation). The back end handles global consistency: it stores poses in a graph and optimises them so the whole trajectory fits together.
Small errors accumulate along the path, a problem called drift. When the scanner returns to a place it has already seen, the algorithm detects this loop closure and spreads the correction back along the trajectory. Post-processing typically repeats this optimisation more thoroughly than in real time, which is why post-processed SLAM data is usually cleaner than the live preview.
Lidar SLAM vs visual SLAM
SLAM can be driven by lidar, cameras or both. Lidar SLAM dominates professional surveying; visual SLAM is common in robotics, phones and consumer devices.
| Factor | Lidar SLAM | Visual SLAM (vision SLAM) |
|---|---|---|
| Sensor | Spinning lidar plus IMU | One or more cameras, often with IMU |
| Measures distance | Directly, with laser ranging | Indirectly, from image geometry or depth cameras |
| Lighting | Works in darkness | Needs adequate light and texture |
| Typical accuracy | Centimetre level for survey devices | Variable; scale can drift without depth data |
| Cost | Higher | Lower |
| Main use | Surveying, mining, construction, forestry | Robot navigation, augmented reality, consumer scanning |
Many survey SLAM scanners add cameras to colourise the cloud, and some use visual features to help in geometrically repetitive spaces.
Types of mobile mapping systems
Mobile mapping systems differ by platform, and each balances coverage speed, accessibility and positioning method.
- Handheld SLAM scanners: light, single-operator devices for buildings, stairwells, plant rooms and underground spaces.
- Backpack systems: heavier sensors carried on the back, often with GNSS for outdoor work over longer routes.
- Vehicle-mounted mobile mapping lidar: sensors on cars or rail trolleys, positioned by GNSS and IMU, for roads and rail corridors.
- Robotic platforms: robotics SLAM on wheeled or legged robots for repeat inspections and hazardous areas.
Outdoors, a lidar mobile mapping system can combine SLAM with GNSS, which limits drift over long distances. Indoors and underground, SLAM is the only positioning method and control points become more important.
How accurate is a SLAM scanner?
Survey-grade SLAM scanners typically achieve centimetre-level relative accuracy, while absolute accuracy over a large site depends on drift control and ground control.
Manufacturers often quote point-cloud thickness or relative accuracy, which describes how tightly points sit on a flat surface. This is not the same as global accuracy across a building. Absolute errors grow with route length, poor geometry and missing loop closures.
Factors that affect SLAM accuracy include:
- Route length between loop closures.
- Featureless or repetitive spaces such as long corridors and tunnels.
- Fast turns, shaking or very slow movement through doorways.
- Moving objects such as people, vehicles and machinery.
- Number and distribution of control points used to constrain the trajectory.
Where tighter accuracy is needed, combine SLAM with terrestrial scans or survey control. Our article on terrestrial laser scanning accuracy and workflow explains the static alternative.
Handheld SLAM scanning field workflow
Good SLAM data depends mostly on the route. A typical field workflow for a handheld scanner is:
- Plan the route: design closed loops of moderate length that start and end at the same point, and avoid long dead ends.
- Set control: place targets or coordinated points at intervals if the cloud must match a survey grid.
- Initialise: start in an open, feature-rich area and hold the scanner still for the time the device specifies.
- Walk steadily: keep a normal, even pace, turn slowly and pause briefly when passing through doors.
- Close loops: return through areas already scanned, including the start point at the end.
- Review live data: watch the real-time view for gaps and misalignment, and rescan problem areas.
- Post-process: run optimisation, apply control and export to standard formats for point cloud processing.
SLAM vs terrestrial scanning: when to use each
Use SLAM when speed and coverage matter more than millimetre accuracy, and terrestrial scanning when tight tolerances or long ranges matter most.
SLAM suits floor plans, space planning, volume checks, mining and tunnel surveys, forestry inventories and progress capture. It also reaches places where tripods are impractical, such as narrow stairwells and ceiling voids. Terrestrial scanning remains the choice for structural steel, pipework fabrication, deformation monitoring and facades over long distances.
| Need | Better fit |
|---|---|
| Floor plans and space management for a large building | SLAM |
| Steel connections or pipe spools for fabrication | Terrestrial scanner |
| Underground mine or tunnel profile | SLAM, with control at intervals |
| Facade or structure monitored over time | Terrestrial scanner |
| Weekly construction progress capture | SLAM |
Many teams use both: SLAM for overall coverage, terrestrial scans for critical zones, all registered together.
SLAM scanners from Mars Geosystems
The Mars Geosystems handheld SLAM laser scanner is a walk-and-scan device with a 0.1–70 m range at 80 % reflectivity and a 905 nm laser. Point-cloud thickness is 2 cm or less in real time and 1 cm or less after post-processing. The host weighs 925 g, or 1450 g with handle and base, and connects over WLAN and USB.
For projects that need tripod-level detail in critical areas, it pairs with our terrestrial phase-shift scanner. Orders start at one unit, with OEM options for distributors. Send your typical project types through the quote page for advice on the right SLAM scanner setup.
Frequently asked questions
What does SLAM stand for in mapping?
SLAM stands for simultaneous localisation and mapping. It is a technique in which a moving sensor builds a map of its surroundings while also calculating its own position in that map. It is used in robotics, autonomous vehicles, augmented reality and handheld laser scanners that survey without GNSS or tripods.
What is the difference between SLAM and mobile mapping?
Mobile mapping is the general term for capturing spatial data from a moving platform such as a vehicle, backpack or handheld device. SLAM is one positioning technique used within mobile mapping. Vehicle systems often rely mainly on GNSS and IMU, while handheld and indoor systems rely on SLAM.
How accurate is SLAM mapping?
Survey-grade lidar SLAM scanners typically reach centimetre-level relative accuracy, with tightly defined surfaces after post-processing. Absolute accuracy across a large site depends on route design, loop closures and control points. Long routes without loops, featureless corridors and moving objects all increase drift and reduce accuracy.
Can SLAM scanners work outdoors?
Yes. Handheld SLAM scanners work outdoors in streets, forests, quarries and building exteriors, provided there are enough surfaces in range for scan matching. Large open fields with few features can be difficult. Some systems combine SLAM with GNSS outdoors to limit drift over long distances.
What is visual SLAM?
Visual SLAM uses cameras instead of, or alongside, lidar to track motion and build a map from image features. It is cheaper and lighter, which suits robots and phones, but depends on light and surface texture. Without depth sensors, its scale can drift, so survey work usually relies on lidar SLAM.
What software is used for SLAM scanning?
SLAM scanners usually come with processing software that optimises the trajectory, applies loop closures and control, colourises points and exports the cloud. After export to open formats such as LAS or E57, the data can be used in any point cloud, CAD, BIM or GIS application for drafting, modelling, volume calculation or analysis.
Do I need control points for SLAM scanning?
For relative measurements inside a single building, good loop closures may be enough. When the point cloud must match a survey coordinate system, or the route is long, control points measured with a total station or GNSS constrain drift and allow independent accuracy checks. They are strongly recommended for professional deliverables.
Need equipment for this work?
Tell us the job, the country and the quantity. An engineer replies with a configuration and price within one business day. MOQ is one unit; OEM is available.
