SLAM100
Fieldwork Guidebook
SLAM100 Fieldwork Guide
Working Precaution
1. The gimbal between the scanner body and the laser head is a precision
mechanism, please handle it with care and avoid bump!!!
2. After use, please store it in the carry box as soon as possible.
3. The dark part of the laser head is the emission area, so please avoid friction
and collision between this module and hard objects.
4. When taking the equipment out from the box, please use two hands with one
hand holding the handle and the other holding the laser but don’t tough the
scan area, and do not drag it out forcefully.
5. When installing a memory card, please make sure the memory card is
unlocked and there is enough space for acquisition.
6. Please make the electrodes right when mounting the batteries, and then
tighten the screw on the handle.
7. The battery should be placed in the carrying box during transportation to
avoid contact with liquids or bumps with hard objects.
8. Please pay attention to the direction of the positive and negative electrodes
when charging and use.
9. Be careful not to damage the battery from falling after flipping two clips aside
to take them out.
10. During the data collection process, please keep the scanner in front of your
body, in the same direction as you are walking, with the laser head facing
upwards and don’t disturb its rotating with touch or other movements.
11. There are three groups of camera lenses on the side of the fuselage, which
should avoid friction or collision with hard objects.
12. Power on
[Link] light will be solid green after long-press for about 3s and after about
40s, the laser will start rotating and standby.
[Link] light will flash after a short-press which means the device is start to
collect data. Please stay still for about 10s before moving forward so
SLAM100 can acquire environmental data around the start point.
[Link] finishing the acquisition, short press the button so the equipment
stops working and the green light will be on.
[Link] power off the device, the operator needs to long press for 3s till the
green light turns blue and the laser stops rotating.
[Link] pay attention to the level of the battery. The yellow light will be
on when the battery is low, so please charge it after the operation.
13. Please read the user manual and fieldwork guidebook carefully before use.
01
SLAM100 Fieldwork Guide
Fieldwork Guide
1. Scenario Applicability
Since SLAM100 can work without GNSS signal, it can be used in under/over-
ground and indoor and outdoor scenarios, such as underground garages, narrow
passages between tall buildings, etc.; Since LiDAR can work without light, so the
use of Lidar-based SLAM100 is not affected by light, as a result, SLAM100 can
work in Karst Cave, mine tunnels, air raid shelters and other scenes.
Because LiDAR can be affected by refraction and reflection, so the laser-based
SLAM100 are not suitable in scenes with a lot of glasses or mirrors, such as mirror
mazes, transparent buildings, etc.
Since laser-based SLAM100 needs to do matching relying on the pre-acquired 3D
point cloud data, so the accuracy cannot be guaranteed if the map formed by the
data have large area with similar structure, such as long straight corridors with no
geometric differences or squares without any geometric structure, etc.
Mirror Maze Transparent Building
Straight Long Corridor Wide Square
Fig1.1 Scenarios in which SLAM100 has low applicability
02
SLAM100 Fieldwork Guide
2. Division of Survey Area
In use of SLAM100, in principle, as long as it has enough power, storage space,
and there are enough close-loops in the acquisition process, the acquisition time is
not limited by the duration. However, if the area to be collected is very large, it is
suggested to divide it in to blocks to collect data and there should be enough over-
lapping areas between each block, which is helpful for the software doing calcula-
tion.
Fig2.1 is a map of a campus divided into different survey areas. Each color
represents an independent operation. There should be overlapping areas between
operations as much as possible and it is recommended that the overlapping area
should not be less than 1/4 of the survey area.
Fig2.1 Schematic diagram of the division of the survey area
[Link] Planning
SLAM100 is a handheld laser scanner, which needs to be hand-held for data
acquisition, so it doesn’t support automatic route planning. Before the operation,
following contents can be referred to to plan the survey route. Reasonable route
planning can effectively guarantee the overall accuracy of the data.
03
SLAM100 Fieldwork Guide
[Link]-loop Route
In order to reduce accumulated error during the acquisition, it is necessary to plan
closed loops as much as possible, i. e. collect data of the same area in different
time periods. The key of close-loop is to have overlapping regions with length
about 5~10m. The overlapping areas can be coincident in the same direction or in
the opposite direction.
Beginning and Ending
Beginning and Ending of the Close-loop
Beginning and Ending
of the Close-loop
of the Close-loop 5~10m
Wrong Close-loop Right Close-loop
Fig3.1.1 Different Planning of the Close-loop
[Link]-loop in Single-story Buildings
The error will accumulate as the distance increases. For single-story building, there
should be as many O-shaped close-loops as possible instead of linear close-loops,
which may control the error to the minimum range.
Why are multiple small-O-shaped close-loops more beneficial for global accuracy?
Fig3.2.1 are three route plannings of the same area. Measure the shortest distance
between two points and estimate the corresponding error, assuming that every 100m
distance will form a relative error of 1cm.
Distance between Relative Error of Distance between Relative Error of
A-C A-C B-C B-C
L-shape Route 500m 5cm 250m 2.5cm
Single-O-shape Route 120m 1.2cm 250m 2.5cm
Multi-O-shape Route 120m 1.2cm 75m 0.75cm
04
SLAM100 Fieldwork Guide
A A A
C C C
B B B
Round-trip Route Single-O-shape Close-loop Multi-O-shape Close-loop
( Less Effective) ( Better Effective) ( Most Effective)
Fig3.2.1 Route Planning in Single-story Building
[Link]-loop in Multi-story Buildings
In the measurement of multi-story buildings, besides close-loop as many as possi-
ble in each floor, it is also necessary to plan the close-loop between the floors. If
possible, the close-loop should begin and end from the stairs on both sides.
[Link] Distance
The laser ranging range of SLAM100 is 0.5m~120m.
[Link] Working Distance: point cloud data cannot be obtained within 0.5m of
the laser, so please walking in the middle area of the working environment as much
as possible to avoid missing data of front and two sides in small space.
[Link] Farthest Working Distance: The measurement capability of the laser is
about 120m when the reflectivity is higher than 60%. Since SLAM algorithm
requires dense point clouds for calculating, although the data can be obtained at
the farthest distance, its density cannot be used by the SLAM algorithm to calcu-
late. When collecting data on the ground, please keep the object within 50m.
05
SLAM100 Fieldwork Guide
[Link] Method
When collecting data in a wide space(such as underground garage, residential
area, park, indoor, etc.), please hold the device as shows in fig5.1, to collect point
cloud and optical image synchronously.
Ceiling
Forward
Ground
Fig5.1 Handheld Method
Before collecting data in places which is relatively close(such as underground air
raid shelters, tunnels, sewage pipes, etc.), the operator can hold the SLAM100 in
normal position for about 30s, and then slowly change to the horizontal place
(Fig5.2) to obtain more uniform and comprehensive 3D data. But please do not use
this method when collecting texture information.
Ceiling
Forward
Ground
Fig5.2 Horizontally Hold in Long Narrow Space
When using SLAM100 handheld, it is necessary to keep geometric features within
50m in front of the device as much as possible to help matching and improve the
accuracy of the SLAM algorithm. If the geometric features are only on one side in
the environment, it may lead to errors in accumulation and the data accuracy
cannot be guaranteed. In this case, the operator can hold SLAM100 with an
angle(about 45°) against the geometric feature for data collection.
06
SLAM100 Fieldwork Guide
~45°
Geometric Features Geometric Features at Front Form an Angle With The
Only On One Side Geometric Features
Top View of the position with the Geometric Features
Residential Area:
1)If there are buildings on both sides and at front, the operator can hold the device in
normal position and go straight, and the data of the left, front and right buildings can be
collected at the same time for matching;
2)If there are buildings on the left of the device but no building on the right or at front, then
please hold the device with an angle about 45°with the building and walk along with the
building. The device will focus on collecting buildings on the left in this scenario.
[Link] Switching
The basic logic of SLAM algorithm is to match the 3D point cloud of each frame
with the historical frames to calculate the current position and attitude. As a result,
it is necessary to ensure that the current frame and the point cloud of the historical
frame have enough overlapping data, which means the operator needs to collect
enough links when switching scenes in some special spaces to tell SLAM that this
two scenes are in the same coordinate system. Such as two-room switching,
indoor/outdoor switch, corner transitions, etc.
Here are a few common-scene switches and the precautions for data collecting.
Other scenarios can be learned from those examples.
[Link]-room Switch
When collecting indoor 3D data, the room scene is often switched. Before starting
the acquisition, all the doors should be opened as much as possible. If it cannot be
opened in advance, when passes the door, the operator should open the door with
his back to avoid the continuously changing point cloud of the door due to its posi-
tion changing interferes with the matching of the point cloud.
07
SLAM100 Fieldwork Guide
When passing through a door, try to avoid going directly from one room to another,
especially if the door is small. Because sudden space switching may cause the 3D
data of two rooms to be unable to connect; the correct approach should be to
collect as much data as possible of two rooms at the same time when the operator
standing in the middle of one of the room, to help matching two rooms(for example,
a side stand at the door).
When passing through the door, please pay attention to the distance between the
door and the wall to the laser due to the minimum measurement distance of the
laser, which will affect the calculation.
Room A
Room B
Red Line represents areas that have been Green Line represents areas that The green area is difficult to match
scanned in Room A and Room B,Red Dots have been scanned in Room A and Room with the red area with two steps above
representareas in Room A that have been scanned. B after entering Room B
Fig6.1.1 Move Directly From Room A to Room B(Not Recommend)
Red Line represents areas that have been Yellow Line represents areas that Green Line represents areas in Yellow area is a transition, it can match with
scanned in Room A and Room B, have been scanned in Room A and Room B that have been scanned. the red area and also the green area
Red Dots represent areas in Room Room B,Yellow Wavy Line represents areas
A that have been scanned. in Room B that have been scanned.
Fig6.1.2 Scan the room at the door before
[Link]/outdoor Switch
The transition between indoors and outdoors is mainly aimed at the scenes with no
geometric features outdoors, which needs special attention, such as going into an
small empty square. Similar to the transition between two rooms, if you go out
directly, the indoor and outdoor data may not be perfectly matched due to the lack
of link points.
08
SLAM100 Fieldwork Guide
Indoor Indoor
Outdoor Outdoor
Red Part represents areas that have Green Part represents areas that have There is no overlap between indoor and
been scanned inside and outside been scanned outside the room after going out outdoor areas so they cannot be matched
the room,Red Dots represent areas inside
the room that have been scanned.
Fig6.2.1 Going Out Directly(Not Recommend)
Indoor Indoor
Outdoor Outdoor
Red Part represents areas that have been Green part represents areas that have Green area can match well with
scanned inside and outside the room, been scanned inside and outside scanned area in red and outdoor
Red Dots represent areas inside the the room after walking out backward area can be extended according
room that have been scanned. to outdoor scanned area
Figure6.2.2 Walk backward from indoor to outdoor, adding overlap areas(Recommend)
[Link] Point
There are two ways to collect control point. One is to directly use the control point
collection module on SLAM100 for collection, and the position information of the
control point is obtained during the data collection process; the other is to manually
select control points from the output data, using post-processing. The comparison
of two methods is as follows.
Control point collection module on SLAM100 Select when post-processing
Processing Mode Rigid Transformation/SLAM Algorithm Rigid Transformation
Setting position On the ground Unlimited
Coordinate conversion,
Functions precision constraints, data stitching Coordinate conversion, data stitching
Algorithm Support SLAM100 software Third party software
Collect Method Automatic Manual
09
SLAM100 Fieldwork Guide
[Link] Point Setting
Using control points, the operator can obtain absolute coordinates, carry on SLAM
matching control, and data stitching. For different applications, the setting methods
and requirements are slightly different.
[Link] Absolute Coordinates
To obtain the absolute coordinates of the survey zone, the control points should be
evenly distributed and no less than 3. Theoretically, the more control points are, the
more accurate the result will be.
[Link] Matching Control
As to SLAM matching control, the control points are mainly used to correct the drift
data in the SLAM algorithm matching, and the absolute position is added for error
clearing and position control. There should be no less than 3 control points, distrib-
uted in areas that are prone to drift (such as intersection areas when scene switch-
ing, areas with weak environmental geometry features, long straight corridors or
tunnels, etc.). When used for SLAM matching control, it is recommended to set
control points in area with strong texture every 100m, and set more in area with
weak texture.
[Link] Stitching
When dividing survey area, the common control points can be used for data stitch-
ing. It is recommended to set the control points in the middle block, so that the
whole survey area can be connected based on the middle survey area; if it is a strip
type survey area , the control points are distributed in the overlapping area of both
ends. Each overlapping area requires 3-4 control points, as shown in Fig 7.1.1.
Fig7.1.1 Control Point setting for Data Stitching
10
SLAM100 Fieldwork Guide
[Link] Point Collection
Please refer to the product manual for control point collection.
[Link] Considerations
[Link] Speed
Data collection is carried out at normal or slower walking speed in conventional
areas. In narrow spaces, such as karst caves, stairs, and tunnels, the turning
speed should be as slow as possible, and the speed should be slower when
switching scenes, and please ensure that there are as much space as possible in
front of the nose.
[Link] Objects Avoidance
SLAM matching needs to match static data, so please don’t face toward scene with
lots of moving objects or people as far as possible. If a large range of the collected
data is moving objects, when the matching is performed based on the moving
object, it will cause the displacement of the static objects. When encountering such
a situation, try to avoid it as much as possible. When it cannot be avoided, stop
walking until the moving object leaves. Here are some situations needed to be
avoided.
[Link] Facing Moving Crowds
During data collection, it is necessary to avoid the entourage in front of the opera-
tor, who should walk behind the operator or not follow; avoid data collection in
crowd.
[Link] Facing Large Moving Objects
When collecting data from an open-pit mine, avoid mining trucks driving at a close
distance (within 10m) ahead; avoid collecting data near large vehicles such as
buses and passenger cars.
11
