AEROBO marker & AEROBO cloud Reference Point Surveying – Technical Report

Aerosense offers single-frequency GNSS survey instrument called “AEROBO marker” and an associated cloud service called “AEROBO cloud”. Together, the system provides high-precision positioning calculations for a variety of survey applications.

Once the GNSS log data obtained by the AEROBO markers is uploaded to AEROBO cloud, a series of three-dimensional network average calculations required for reference point surveying is automatically executed (automated methods are patent pending). In this report, we introduce the accuracy and the level of efficiency gained, using actual reference point survey as an example.

Conventional methods vs. AEROBO marker & AEROBO cloud

In the general construction workflow, a detailed drawing of the site is created before the construction starts, and a reference point is established. It is necessary to carry out a reference point survey to comply with local building regulations. The survey confirms high grade 1st or 2nd class reference points on the site from which 3rd and 4th class reference points are established. In the case of a large site, there are likely to be many new reference points to be established, typically involving multi-network observation scans. This series of operations related to the reference point survey has become a burden for contractors.

AEROBO marker is a GNSS logger equipped with GNSS receiver. It is lightweight and has the advantage of ease of operation. In December 2019, the Geospatial Information Authority of Japan approved the registration of the instrument, and it is now available as a single frequency GNSS survey device for all reference point surveys, including mapping. (

AEROBO markers are installed in the field in accordance with the survey plan, and observations are carried out for more than one hour. Since the operation requires nothing more than the pressing of the power button on the AEROBO marker, it can be carried out by anyone in the field. If necessary, a tripod and a levelling table can be used in cases where the AEROBO markers cannot be located directly on the ground, such as on survey piles.

Once the on-site observation is complete, the observation log is uploaded to AEROBO cloud. Once uploaded, the following sequence is required for the reference point survey to be performed automatically:

  1. Multi-session division of observation log.
  2. Generation of polygonal nets.
  3. Baseline analysis.
  4. Inspection calculation.
  5. 3D network average calculation (hypothetical network, exact network)
  6. Report outputs in accordance with the work regulations.

In the example below, the actual reference point survey shows a case of installing 25 new points on the site from the 2nd class reference point that are 3 points around (electronic reference points, 301, 302, and 303).

A closed polygonal network composed of 3 known points, 25 new points, and 73 baselines.

Using conventional methods and a total station to survey 25 new points would typically involve two people on site for one day. The subsequent post processing work using net average software would take some 3 hours, plus a further 2 hours for the final calculation. A total of about 14 hours work.

Using AEROBO markers and AEROBO cloud, we were able to complete the same job in 6 hours, representing 60% increase in efficiency.

Unlike a total station, AEROBO markers can be measured simultaneously. Meaning that the more observation points, the greater the level of efficiency of operation. Furthermore, the fact that calculation can begin as soon as the log has been uploaded greatly contributes to the overall time saving.

Fully automatic three-dimensional network average calculation and calculation accuracy

To verify the accuracy of AEROBO marker & AEROBO cloud 3D network calculations, we used reference points with public survey results. From the online reference point database provided by the Geospatial Information Authority of Japan, the survey year and 9 reference points with different reference point grades are comprehensively extracted (see figure below). The location was measured with AEROBO markers. For simplicity, the marker was placed directly on top of the reference point. The difference between the original value and that obtained using the AEROBO marker measurement value is evaluated as an error.

Public survey reference point used for verification – Hitachinaka City

AEROBO cloud uses a single frequency static positioning system, GPS satellite L1 signals for baseline analysis (*). Static positioning requires the observation log and the survey reference point (a given point, corresponding to a known point) observation log. Then, AEROBO cloud automatically searches the nearest 3 electronic reference points around the site. At the verification site, 3 electronic reference points were selected: Hitachi, Mito, and Koda.
(*) Scheduled multi-GNSS support, such as GLOANSS and QZSS, during fiscal 2020

In the Hitachinaka example, the observation points are distributed far away in between. Since the distance between observation points is too long as the baseline between new points (*), we considered a combined polygonal network for accuracy evaluation, consisting of one new point and three electrical reference points for each observation point (the figure as below). We will later describe the polygonal network of 3rd and 4th class reference points normally carried out on site.
(*) The distance between new points at 1stclass reference point is 1km. As the grade decreases, the new point interval decreases.

In addition, to confirm the effect of the three-dimensional network average calculation, static positioning calculation using only the nearest electronic reference point, Mito (no network calculation) was also performed together.

The results are summarized below.

A combined polygonal network consisting of one new point and three
points. The known points consist of electronic reference points, 301 (Hitachi), 302 (Mito), and 303 (Koda).

The baseline length is approximately 18km for Hitachi, 10km in Mito, and 20km in Koda.

Table 2.

Baseline analysis results connecting one nearest electronic reference point and each observation point.
No net calculation.
The observation point number corresponds to the observation point number described in Table 1.
The error represents the size in meters.

Table 3.

Combined polygonal three-dimensional network average calculation results composed of three nearelectronic reference points and each observation point.
The observation point number corresponds to the observation point number described in Table 1.
The error represents the size in meters.

Scattered plot notation of the results of Table 2 and 3.
Horizontal axis: Horizontal error, Vertical axis: Vertical error. Both axes are in meters. The right figure shows an enlarged range of the left-hand horizontal axis 0.03 and the vertical axis 0.1. In the figure, the blue point shows the result without the net calculation (Table 2), and the orange point shows the result with the net calculation (Table 3). The number described next to the point represents the observation point number, corresponding to the observation point number described in Table 1.

Table 2 and Table 3 shows coordinate value X, Y, Z in the plane right-angle coordinate system 9 in the cases of “with network calculation” and “without network calculation”. Furthermore, the difference ΔZ from the survey result value in the database is described as a vertical error, and the error size calculated from ΔX, ΔY is described as a horizontal error. The figure above shows the distribution diagram of the horizontal error vs. the vertical error. In this plot, point 9 was excluded since the fix rate is low (less than 10%), rendering the reliability of the value low. (The reception of radio waves at Point 9 is presumed to have been deteriorated due to its location at the base of the large tree.)

From the above figure, the observation point error is distributed in two regions: 10cm or less (in the red frame), and 20cm or more (outside the red frame). In the error range of 10cm or less, there are 5 points in 9 observation points. Without network calculation, their horizontal error average is 1.3cm, and their vertical error average 3.6cm. On the other hand, with net calculation, the horizontal error average is 1.1cm, and the vertical error average 0.96cm. By performing the network calculation, the accuracy was improved in both directions, especially the vertical accuracy was greatly improved.

In the error range 20cm or more, there are three points in nine observation points. The common point is that the survey year of the result value is before 2011. The following figure shows a diagram plotting the error size (no net calculation) in the vertical axis and the horizontal axis: survey year. As the reference point of the survey year is older, the error between the positioning result and the electronic reference point is greater. In particular, the trend is remarkable before the survey year 2011.

The survey results of the electronic reference point are the values at the original value (2011 survey results), and the crustal movement amount parameter for converting to the original value is updated annually (so-called semi-dynamic correction). On the other hand, the reference point derived from the triangular point registered in the reference point database is based on the measurement value of the year surveyed, corrected to the original period value, and then fixed. Therefore, the reference point derived from the triangular point diverges from the result value derived from the electronic reference point each year. (To begin with, the old triangular points are often in the field, so the reference points are often moved.)

In the site where its survey is conducted based on the old triangular point, it is desirable to do GNSS surveying based on the old triangular point as a reference point. AEROBO cloud also supports static positioning, using the known point as a reference point. For example, when the point 4 is used as a reference point to calculate the point 5 of the same survey year, the horizontal error is 11cm, and the vertical error 2.2cm. Therefore, the accuracy is greatly improved from the vertical error 48.3cm and the horizontal error 19.2cm in the case of the electronic reference point (Table 2). (Table below)

In general, when performing GNSS survey with a reference point, it is better to use an existing reference point on the site, as it is possible to obtain a survey accuracy that is consistent with the site coordinate system.

Since it takes time to set up markers when the number of observation points increases, it is necessary to perform long-term observations of more than one hour, or to perform multi-session observations that divide observations into multiple parts, so that the observation time between observation points (between new points) overlaps. AEROBO marker & AEROBO cloud supports long-term observation and multi-session observation. For example, 25 new points introduced in the previous part was calculated in three sessions as below.

An example of a 3D multi-dimensional network automatically generated by AEROBO cloud. It consists of 25 new points and three known points, equivalent to almost the maximum calculation score that can be handled by AEROBO cloud. The analysis accuracy of each baseline is color-coded: in the figure, the blue point/ baseline is within the specified precision, and the red point/ baseline is outside the specified precision).

A closed polygonal network of the 73 baselines with each new point being the vertex of the triangular mesh is automatically generated, and the net calculation result was the following.

Even though the computing cost of 73 baselines and their network calculation is normally very high with conventional methods, AEROBO cloud can process them cheaply and automatically, and the deviation of the oblique distance of each baseline vector and the standard deviation of the new point are within the allowable range of the 4th class reference point defined by the reference point surveying work provisions.

Compatible with any GNSS surveying equipment

In October 2019, AEROBO cloud started to support any GNSS surveying equipment other than AEROBO markers and is now able to measure reference points where the nearest electronic reference point is more than 10 km away from the survey site. (That is, it is now possible to do all reference point surveys.)

For this specific example, there is a detailed report of various verification results using HiTarget Surveying Instrument Co.Ltd’s dual frequency GNSS receiver by Koizumi Sokki Manufacturing Co., Ltd. Please refer to the following URL:

As of March 2020, AEROBO cloud supports 80% of the report formats that common commercial network calculation software outputs (all major formats are already supported). A list of printable reports is listed at the end of this report. These reports can be submitted to government agencies and other agencies in accordance with the standards and regulations of the reference point surveying work.


  • AEROBO marker & cloud was found to reduce all work hours by 60% compared to traditional total station/ general commercial software. Since AEROBO markers enable multi-point simultaneous measurement, the more the observation points, the larger the reduction effect.
  • The three-dimensional network average calculation accuracy of AEROBO cloud was evaluated using a public survey reference point whose result value was known. As a result, the horizontal error average was 1.1cm and the vertical error average was 0.96cm. This is sufficiently clear for the single frequency GNSS surveying equipment performance standard of AEROBO marker & cloud (the shortest baseline length of 10km).
  • Even 25 new points, 3 known points, and their large-scale polygonal network calculation consisting of 73 baselines, AEROBO cloud can process. The deviation of the resulting oblique distance and the standard deviation of the new points are within the allowable range of 4th class reference point.
  • In AEROBO cloud, regardless of the size of the dataset, once the log is uploaded and the calculation is started, anyone can do the reference point surveying. Furthermore, since the electronic reference point is supported as standard, even if there is no reference point in the field, it is possible to do the reference point surveying.

Compared to AEROBO cloud, when using a general commercial network calculation software, the degree of freedom is high in terms of a series of reference point surveying processing such as the adjustment function of report layout and the creation of the polygonal network of arbitrary shape. However, it requires editing the polygonal network manually, such as the examination of the calculation results for each step of the workflow. There are a lot of user burdens in a general commercial network calculation software.

Comparison table of common commercial network calculation software and AEROBO cloud

With AEROBO marker & cloud, even without knowing the details of the reference point survey, anyone can easily conduct a 3D network average calculation and create the required survey reports. Because it has sufficient performance / calculation accuracy, if the survey plan is set up appropriately in advance, it can be an effective tool for utilizing human resources and reducing survey costs.

Supplement 1. Supported reports

Supplement 2. Antenna Phase Characteristic Table of AEROBO marker

AEROBO cloud, in the baseline analysis, performs antenna phase correction. The antenna phase correction table of AEROBO marker used for verification is as follows. The table has been approved by Geospatial Information Authority of Japan. (15 October 2019 version)