The objective of this project is to generate an algorithm (to be run on a mini pc) to create a cloud of data points obtained from 3 systems: LIDAR, GPS RTK and IMU. For this the following aspects must be taken into account:
When it comes to the point cloud map creation, we need to know exact position of each scanned point in the ﬁnal coordinate system bounded to a local ground. To this point, we were only able to acquire coordinates relatively to a LiDAR’s position. Using an integration of an IMU and a GPS module.
Let’s assume that the communication lines have a negligible traﬃc delay and the processor of the on-board computer receives the data values almost immediately after they were measured by the sensors. The principle is the same for each of the three devices. Right after the data is read from the line by a dedicated method, the computer’s current system time is assigned to the record. The record is then stored into a data buﬀer. Each sensor has its own buﬀer and a thread managing the sensor’s services.
.We take a LiDAR’s point from its buﬀer.
.We take ﬁrst two IMU’s records from its buffer.
.If the LiDARs point time stamp is lower number than the time stamp
of the ﬁrst IMU record, the point is discarded. This repeats until the
point’s time stamp is greater than the ﬁrst IMU record’s time stamp.
.When the point’s time stamp lies between the two time stamps from IMU
records, the two IMU records are used for the point’s transformation.
.If the point’s time value is greater than the second IMU record’s time,
the ﬁrst IMU record is discarded, the second takes place as the ﬁrst and
the second place is ﬁlled with a new record from the buﬀer.
We described the principle on IMU records but the same applies to the
GPS records in parallel.
Make coordinates systems transformations
Calculated point cloud with raw sensors records
This approach stores raw data from each sensor separately resulting in three
binary ﬁles as an output from the vehicle after a survey. A point cloud from
it is calculated on a PC and allows for better debugging. It also gives us the
possibility to generate models with various parameters from one data set. If
we detect some major problem in the data set, it can be ﬁxed.
Offline conversion to LAS Standard
Correction Imprecision and Uncertainty Problems