User's Manual

STONEX SLAM XH120 – User Manual
39
7. Operational recommendations
Before scanning, user should plan the optimal scanning path to identify potential problem areas, such as
harsh environments, through doorways and stairwells, etc. The scan path of the plan should also be a
"closed loop" as much as possible. Please refer to the guidelines of these chapter in order to achieve the
best scanning results.
7.1 Environment
The SLAM algorithm, which is used to process raw laser scan data into a 3D point cloud, relies on
identifying features present in the scanning environment, which are scanned repeatedly when the operator
passes through the scanning environment. For the feature to be prominent, the ratio of its size to the
range must be approximately 1:10. For example, within the range of 5m, the size of objects with distinctive
features must be greater than 0.5m.
Environments with "poor characteristics" include open spaces (such as roads without buildings, smooth
long tunnels, etc.). In these environments, the SLAM algorithm may not have enough features in the
direction of travel to determine the moving direction. We recommend the following steps:
1. If possible, use other objects to enhance the environmental characteristics. For example, in an
empty long corridor, placing boxes or parking vehicles in open spaces, etc.
2. When moving in the environment, please ensure that any available features are scanned
repeatedly. By doing so, enough feature points can be obtained, which increases the possibility
of using this feature by the SLAM algorithm. This is especially important when the element is at
a long distance (>10m). For example, when scanning a smooth wall passage, the only feature in
the direction of travel is the end of the wall passage or door.
3. Avoid scanning moving objects (for example, pedestrians or vehicles walking close to you during
scanning, especially large trucks), because the SLAM algorithm may lock these objects as static
features, which will cause the mapping to fail.
7.2 Loop closure operation
The SLAM algorithm uses the previous positions to determine its current position. This method may
introduce some errors and cause the measurement position to "drift". A good measurement habit is to
"close the loop" by re-visiting a known position so that the compound error can be constrained within the
loop range.
The minimum requirement for loop closure operation is that the operator must start and end the
measurement at the same location to ensure that at least one loop is closed. However, it is recommended
that the operator close the loop as often as possible, to minimize errors and improve the accuracy of the
resulting point cloud. Replace a big single loop with multiple small loops.
It is also better to make a "circular loop" rather than a "forward/backward loop", because in the latter case,
the survey path simply doubles itself, which is not helpful for the SLAM algorithm. That is, if possible, enter
and exit through different doors, and move between floors through different stairwells.
It is important to scan the closed loop area carefully to ensure that key features are scanned from a similar
angle. If you return to an area from another direction, it may be necessary to turn around. This is especially
important in environments with poor characteristics.