STUDENT CONTEST 2018
“Precise Georeferencing” A Cost-Effective Alternative to Ground Control in an RPAS Survey Andy Kearns, BSc (Hons) Geomatics (Surveying & Mapping) School of Surveying & Construction Management, Dublin Institute of Technology, Ireland.
Summary Traditionally, topographic land surveying has been heavily reliant on the use of total stations and GNSS technology. In more recent years, Remotely Piloted Aircraft Systems (RPAS) have been utilised to acquire topographic information. RPAS surveys generally contain Ground Control Points (GCP’s), which allow the images captured to be georeferenced. Recent advancements in, RPAS contain Real Time Kinematic (RTK) positioning systems, eliminating the need for GCP’s, however, such systems increase the cost involved with RPAS surveys. The aim of this investigation was to determine if the use of a total station in RPAS surveying, would achieve an acceptable topographic surveying standard as defined by the Royal Institute of Chartered Surveyors (RICS) of 0.050m in X, Y and Z, without the need for GCP’s.
A mini prism attached to the RPAS, allowed for the total station to autonomously track and continuously record the RPAS position when each photo was taken. A single flight was performed with ground control in place. The data captured in the single flight was processed in Pix4D mapper pro three times, firstly, with GCP’s, secondly, without GCP’s and finally using the coordinate information obtained from the total station. The accuracy of these three data sets were subsequently determined from the Easting, Northing and Elevation of eleven check points whereby the baseline data for this research was the RPAS survey with GCPS’s.
The results of this investigation were determined using the eleven marker/check points sprayed on the carparks surveys. The Easting, Northing and Elevation of each marker point from the three methods was compared. The base data for this research was the RPAS survey containing GCPS’s. The findings of this research determined that the survey containing GCP’s performed 44% better in the Easting, 8% better in the Northing and 27% better in the Elevation than traditional RPAS methods using ground control points
1 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 However, these findings did not meet the RICS topographic land surveying standards, however, there were several variables, such as camera bias, shutter bias and GPS bias, which could be changed to achieve better results, leaving this method RPAS surveying open to further investigation.
Introduction Remotely Piloted Aircraft Systems (RPAS) are commonly referred to as Drones, Small Unmanned Aircraft, and Unmanned Aerial Vehicles but for clarity here the term RPAS will be used. RPAS technology is increasing at a fast pace, allowing for quicker data acquisition. RPAS can gather high resolution imagery in a fast, efficient and safe manner for the user. With the growing popularity of RPAS among both hobbyists and professionals alike, RPAS technology is in demand. The growing advancements in RPAS technology have led to a rise in manufacturing costs which transfers onto the consumer market.
RPAS come in all shapes and sizes, from the non-professional Dji Spark, to the Trimble UX5 Real Time Kinematic (RTK) positioning RPAS. RTK RPAS have a major benefit to the land surveying industry, allowing for large areas to be surveyed, in a fast and accurate manor without the need for Ground Control Points (GCP’s). For RPAS that do not have RTK capability, GCP’s are required to allow for the RPAS images to be processed in real world coordinates. In accordance with Royal Institute of Charters Surveyors (RICS) Measured Surveys of Land, Buildings and Utilities 3rd edition (RICS, 2014), medium to high accuracy topographic land surveys must be within a maximum plan accuracy of 50mm and a vertical accuracy of 50mm. Currently, this accuracy is only achievable using GCP or high grade RTK systems.
This investigation aims to determine, if integrating a Total Station and RPAS has the potential to achieve similar accuracies to RPAS surveys using GCP’s. This method of integrating the total station with a RPAS will be known as the “Precise Georeferencing” method.
2 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018
Background To operate a RPAS in Ireland, as in many other countries, users must conform to specific legislation and regulations that are set out by the Irish Aviation Authority (IAA) and conform to the guidelines for the operation of model aircraft as set out in the Irish Aviation Authority Small Unmanned Aircraft (Drones) And Rockets Order, 2015 (IAA, 2015).
As discussed by Mian et al., (2015) the cost-effective solution of Multi-Rotary RPAS, such as the DJI Mavic Pro have accuracies of +/-3/5 meters. This does not meet the standards for surveying as defined by the RICS (RICS, 2014). Thus, the use of GCP’s are often integrated into RPAS surveys. Ground control is required to calculate the scale, orientation, and absolute position information of the products for increased accuracy (Micheletti et al., 2015).
Ground Control is usually obtained using Network Real Time Kinematic (NRTK) positioning rovers such as the Trimble R10, which has accuracies of +/- 0.008m X & Y and +/- 0.015m in Z.
3 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018
Direct Georeferencing is a new technique in photogrammetry which does not require ground control and aerial triangulation, to process in a known reference system. An advantage of direct georeferencing is the processing speed resulting in a simpler field and processing workflow (BlĂĄha et al., 2012). Originally, the direct georeferencing method was designed for high accuracy mapping from manned aircraft. This method incorporates high end GPS and Inertial Measurement Units (IMU) to determine the location in which the image was taken. However, more recently this method has been introduced into RPAS surveys, as seen in the Trimble APX fixed-wing RPAS (Figure 1). The Trimble APX operates a small GNSS inertial measurement system that allows the RPAS to operate without the need for ground control (Trimble, 2018).
Figure 1: Trimble APX RPAS (Trimble, 2018)
The direct georeferencing method, involves internal components operating simultaneously to accurately georeferenced the images.
4 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018
Methodology The methodology for this research included several elements: Designing a suitable mount for a 360° prism and determining the offset from the prism to the charged couple device in the drone, identifying a suitable site and capturing and processing the flight data.
1. Stable mount The need for a stable platform to mount the mini prism was a key factor in this study in addition the mount would have to be ultra-lightweight while not interfering with any key components of the RPAS itself. The steps involved in designing a suitable solution are illustrated in Figure 2. On initial analysis of materials, it was found that some metals adversely reacted with the Mavic Pro compass and it was decided that aluminium which consists of ferromagnetic materials was best suited. As aluminium is extremely light weight and strong it satisfied some of the other design criteria.
Design prototype
Define what was needed
Gathering information
Refine
Final Product
Figure 2: Design Workflow
To ensure a stable consistent platform to mount the prism, 3D printed parts for the DJI Mavic Pro were found on the website www.shapeways.com. These parts suited mounting a GoPro Action sport camera to the top of the RPAS. With the 3D printed mount clipping into same place each time, it provided the perfect platform for stability and consistency as it reduced the gravitational forces by spreading the payload around the RPAS. The final design used in the methodology can be seen in Figures 3 & 4. In total, the fabrication of this mount cost an estimated €55 or $65. 5 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018
Figure 3: Fabricated mount
Figure 4: Mount attached to RPAS
6 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 2. Offset Determination To accurately determine the position of the aircraft in the air, an offset from the centre point of the prism to the point of where the image was taken was required. When an RPAS takes an image, there is a determined offset from the charged couple device (CCD) in the camera to the IMU and GPS unite built into the RPAS. Thus, a significant research element included accurately determining where the CCD was in the RPAS. As the CCD is an internal component of the RPAS it was not within the scope of this research to dissemble and reassemble the drone. Therefore, the best approach was to determine the offset from the prism to the centre focal point of the lens of the camera. Various methods were tested to determine this offset, including Structure for Motion (SfM) using a Samsung handheld device and hand-held laser scanning, however these were not sufficiently accurate. The most accurate method adopted was using a total station which precisely recorded distances and angles, from which application of the trimetric cosine rule enabled determination of the height difference between the prism and the camera. An offset between the CCD and the prism of 0.147m was determined, as illustrated in Equation 1.
Equation 1: Offset Formula
đ?‘Ž2 = đ?‘?2 + đ?‘? 2 − 2 đ?‘?đ?‘? ∗ đ??śđ?‘œđ?‘ đ?‘‹ đ?‘Ž2 = (1.5473)2 + (1.5265)2 − 2 1.5473 (1.5265) ∗ đ??śđ?‘œđ?‘ (5°26"30") đ?‘Ž2 = 4.7243 − 4.7026 đ?‘Ž2 = 0.217 đ?‘Ž = 0.147
7 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 3. Site Selection
The test site location used for this research was Whitehall Colmcilles GAA Club in Dublin, Ireland (Figure 5). The southern aspect of the car park and the flat roof located in the top of Figure 1 was used for this research. The test site fell under Irish Aviation Authority (IAA) ‘Class C’ airspace which required a formal permission from the IAA together with a risk assessment. The relatively flat surface of the car park provided a suitable location for acquiring the survey however, obstacles including power lines and adjacent private property restricted the flying height.
Figure 5: Test Site Location
8 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 4. Establishing Control
The initial stage of the fieldwork involved establishing control to enable the total station to operate using real world coordinates. Figure 6 shows the location of the two control points (CP01 and CP02) used for this research, these points were selected on the basis of avoiding obstructions and including varying elevations. CP01 was situated on an open area and doubled as a ground control point for the aerial survey. CP02 was selected based on elevation being 3.5 meters above the ground area which permitted lover tracking angles for the MS60 to track onto the RPAS prism.
Figure 6: Control Location
The two CP’s (CP01,CP02) were surveyed using a Trimble R10 GNSS receiver in Network Real Time Kinematic (NRTK) mode. To obtain a reliable and accurate reference network, each point was surveyed for 180 epochs, which takes a measurement in one. From the coordinates, an RO check was carried out using the Total Station to ensure absolute accuracy of the control stations (Table 1)
Table 1: RO Check
9 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 Seven ground GCP’s were positioned homogenously around the survey area (Figure 7). These ground targets were made up of 350x350mm chess board tiles attached to timber. The targets were positioned using the Total Stations active tracking and a 360° mini prism. To compare the positional accuracy of the RPAS imagery without GCP’s and the RPAS with direct georeferenced to the base data (RPAS survey containing GCP’s) using An additional 11 marker points which were sprayed onto the surface of the survey area. As with the CP’s, the marker points were spread around the survey area, with some to the outer edges to evaluate any distortional effect that may occur (Figure 7).
Figure 7: Ground Control & Marker Point Location
10 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 5. Flight Data For this research, two methods of data capture were required. The RPAS mounted with the 360° mini prim was flown, while the Leica MS60 Total Station autonomously tracked the prism. These components were required to work in synch and without interruption to ensure that each image coordinate from the Total Station could be matched with the image time stamp.
Pix4D mapper pro was a key component in achieving the required data for this research. The application allowed for an autonomous flight to be carried out, allowing the Total Station to be operated at the same time. Pix4D capture was the software used in this research. The application has four modes of operation: Grid mission (best for 2D/3D Mapping), double Grid mission (Best for 3D Modelling), Circular Flight mission (Building Modelling) and Free Flight mission which is for distant imagery. The mode chosen for this research was the grid mission, reasoning for choosing single grid over the double grid was the amount of imagery captured. Based on Barry and Coakley (2013), a flying height of 20m above ground level and at a 90° angle camera nadir imagery with an overlap of 80% was adopted. The RPAS flight took roughly 6 minutes and covered an area of 0.748 Acres. Forty-nine images were captured at roughly 1m intervals along the flight path. The flight plan used can be seen in Figure 8.
Figure 8: Flight Path
The Total Station’s automatic aiming was optimised to continuously track the prism mounted to the RPAS. On site, the Total Station was set up on an adjacent rooftop, at roughly 3.5m 11 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 above the survey area, this allowed lesser angles between the Total Station and the RPAS. The coordinated control point as mention previously allowed for the Total Station to operate in the chosen coordinate reference system, Irish Transverse Mercator (ITM).
The continuous measurement mode of the Total Station was paramount in the investigation. The automatic recording was set to measure a 0.020m change in position. During the 6-minute mission, the Total Station maintained a constant feed with the prism attached to the RPAS. The Total Station recorded 1,620 points for the RPAS survey, with only 49 images taken during the survey.
When acquiring the data, both the RPAS and the Total Station recorded a timestamp to each image or coordinate capture, this was used as the basis for image coordination of the georeferencing. The timestamp on the total station was automatically recognised as GMT +00.00 and could be adjusted in the manual user settings, whereas the RPAS determined time from the device, i.e. a Samsung phone attached using GMT. The timestamp and corresponding image number can be seen in figure 9.
Figure 9: Time stamp matched with image
12 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 6. Data Processing
As the Total Station recorded in increments of change by 0.020m in distance, over 1,600 points were required to match with the 49 images. Thus, there were a lot of redundant measurements and it was apparent that there were varying numbers of measurements for each time stamped image.
The potential adverse effects of the varying number of measurements per image are unknown and remain unknown, so until an exact timing solution between the Total Station and the RPAS can be determined an exact solution for time difference can’t be determined. Therefore, all positional data for the 46 images were averaged, the predetermined offset from of 0.147m from the lens of the RPAS to the centre of the prism was then deducted from the height of each image.
The RPAS processing software used for this research was Pix4D Mapper Pro for in house processing and Pix4D Capture, for acquiring the data. Figure 11, shows how the data was processed. It is clear that there were more steps involved in the traditional method, which resulted in a longer processing time from start to finish. The point cloud generated from the precise georeferencing method can be seen in Figure 10 below.
Figure 10: Point cloud from precise method
13 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018
Figure 11: Processing workflow
14 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018
Analysis To satisfy the research aim, four main datasets where analysed and compared. The eleven check points were used to determine each 3D positional accuracy of each test flight. The Total Station and RPAS with GCP’s were used as the baseline dataset for this analysis and the findings of the study are based on the Mean Error between the baseline and test data sets.
In this investigation, the Total Stations performance on the marker points were used as the gold standard. From Table 2, the Gold standard RMSE was 0.026m in Easting, 0.029m in Northing and 0.088m in Height. The results from Table 2 indicate that the accuracy achieved is of an acceptable RICS standard in the Easting and Northing as set out by the RICS, but does not meet requirements in elevation
Table 2: GCP method
Table 3, lists the results of the RPAS survey processed without using GCP’s. The RPAS data processed without ground control resulted in an RMSE of 1.893m Easting, 0.829m Northing and 42.033m in the height. This outcome was expected as it was not directly tied down to any reference network. The Pix4D software processed the images using the RPAS Exif data which was acquired from the internal measurement system. The accuracy of this dataset is not compatible with most surveying applications, however it could be argued that for 2D boundary surveys, given the +/- 1.5m Ordnance Survey Ireland (OSI) boundary accuracy it is acceptable.
15 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 Table 3: No GCP Method
Table 4 lists the results of the Precise Georeferencing method. The Precise Georeferencing method using the prism mounted to the RPAS being tracked by the Total Station was not as expected. The accuracy achieved was 0.559m Easting, 0.386m in the Northing and 0.319m in Height. The accuracy achieved by this method would be acceptable in some surveying applications such as cliff face surveying where it may not be possible to insert ground control on the side of a cliff or on the rocky shore line. Figure 12, graphically illustrates the RMSE values for each dataset.
Table 4: Precise Georeferencing Method
16 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018
Figure 12: Comparison of results
The results of this research found that the more traditional approach to RPAS surveying using ground control performed 44% better in the Easting, 8% better in the Northing and 27% better in the Height in comparison to precise georeferencing method. It is clear from the results as presented that the expectation of matching or improving the accuracy of RPAS survey by integrating surveying instruments using Precise Georeferencing was not met. Many factors may have impacted on these results including single grid mission, time synchronisation and offset determination. A contributing factor to the poor performance of the precise georeferencing method could be the decision to use the single grid mission instead of the double grid mission. The double grid mission would have given the software more images, tie points and overlap, although the negative side to this would have been the processing of the total station data as the amount of points recorded would be doubled. The time synchronisation may have also been a contributing factor to the poor performance of the precise georeferencing method. The initial time offset between the total station and the RPAS itself was rectified, but the shutter time offset in the RPAS was not known, therefore making it impossible to rule out time bias. In the processing of the total station data a major contributing factor to the result was the RPAS not measuring to the millisecond. A possible 17 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018 solution to overcome this time bias would be to manual perform the free flight mission, stop and record the drone as each image is taken. The RPAS flight parameters may have had an adverse effect on accuracies achieved, these include flying height, direction of flight and the use of a single grid. The RPAS itself may also have adversely influenced the results, a higher grade RPAS could potentially have provided better pixel values resulting in a denser point cloud. The shutter speed and distortion caused by the shutter could also have affected the results.
The interior and exterior camera orientations, known in Pix4D as Omega, Phi and Kappa, may have also have had a negative effect on the accuracy results. When the CSV file containing the RPAS positions was imported into Pix4D, the Omega, Phi and Kappa values were not processed by the software. Pix4D states that “due to different UAV manufacturers it is not possible to guarantee that Omega, Phi, and Kappa can be calculated accurately for all UAVs.� (Pix4D,2017). Thus, the missing interior and exterior orientations values may have had an adverse effect on the accuracy of the precise Georeferencing method. 4A possible solution to acquire more accurate survey data to meet RICS standards would be develop a formula to be integrated into the photogrammetric processing. Such formulae would allow for a value to be assigned to each unknown parameter which may have a positive affect on the results, an example can be seen in Equation 2 below.
Equation 2: Bias Correction
đ?‘Šđ?’Šđ?’‚đ?’” đ?‘Şđ?’?đ?’“đ?’“đ?’†đ?’„đ?’•đ?’Šđ?’?đ?’?đ?’” đ?‘şđ?’‰đ?’–đ?’•đ?’•đ?’†đ?’“ đ?‘şđ?’‘đ?’†đ?’†đ?’… đ?’ƒđ?’Šđ?’‚đ?’” + đ?‘ťđ?’Šđ?’Žđ?’† đ?’ƒđ?’Šđ?’‚đ?’” + đ?‘Žđ?‘ˇđ?‘ş đ?’ƒđ?’Šđ?’‚đ?’” + đ?‘ˇđ?’“đ?’?đ?’„đ?’†đ?’”đ?’”đ?’Šđ?’?đ?’ˆ đ?’ƒđ?’Šđ?’‚đ?’” = đ?‘ťđ?’?đ?’•đ?’‚đ?’? đ?‘Šđ?’Šđ?’‚đ?’” đ?‘ťđ?’?đ?’•đ?’‚đ?’? đ?‘Šđ?’Šđ?’‚đ?’” − đ?‘Şđ?’?đ?’Žđ?’‘đ?’–đ?’•đ?’†đ?’… đ?‘˝đ?’‚đ?’?đ?’–đ?’† = đ?‘ťđ?’“đ?’–đ?’† đ?‘˝đ?’‚đ?’?đ?’–đ?’†
18 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018
Conclusion The findings of this research have not met the requirements of the RCIS medium to high accuracy topographic survey standards (RICS, 2014). The research confirmed that in the absence of PRAS with RTK enabled, the use of GCP’s in an RPAS survey is currently the best solution for high accuracy surveys using consumer grade drones, such as the Dji Mavic and Phantom range. The proposed Precise Georeferencing method, depending on the client’s specifications and requirements may be of sufficient accuracy for the required task. A possible application of the Precise Georeferencing method may be of sufficient accuracy for surveying areas of conservation, such as the Burren in Co. Clare where it may not be possible to put in hard control. Other areas were its possible application could be beneficial would be in cliff face surveying, were ground control is not possible. This research was somewhat constrained by the mid-range resolution of the Dji Mavic Pro 4k camera sensor of 12.35 megapixels. Further studies could investigate the accuracies possible or achievable using an RPAS equipped with a high-resolution camera. The main variable found to have an adverse effect on the results was the single grid mission. To gain more accurate values a double grid mission should be flown. As mentioned, a solution to the time synchronisation difference would have to be determined. Bias from the RPAS camera to the GPS would have to be determined and mitigated to allow for better time stamp matching. Further studies could investigate the achievable accuracies using an RPAS equipped with a high-resolution camera. A possible solution for finding the interior and exterior orientation of the RPAS could be to use the determined Omega Phi and Kappa values as calculated from the Exif data in Pix4D. This would, in theory, allow the software to take into consideration the Pitch, Roll and Yawn of the RPAS while the image was taken. This would possibly allow for better accuracy when processing the data sets. Therefore, there are many parameters of this investigation that could be improved on, leaving this method of RPAS surveying open to further investigation.
19 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24
STUDENT CONTEST 2018
References Barry, P. and Coakley, R. (2013) ‘Field Accuracy Test of RPAS Photogrammetry’, (May), p. 19809. Mian, O, Lutes, J, Lipa, G, Hutton J, Gavelle, E and Borghinic, S. (2015) ‘Direct georeferencing on small unmanned aerial platforms for improved reliability and accuracy of mapping without the need for ground control points’, International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences – ISPRS. Volume XL1/W4, 2015. IAA (2015) ‘Irish Aviation Authority Small Unmanned Aircraft (Drones) and Rockets Order, 2015’, (563). Available at: http://www.irishstatutebook.ie/eli/2000/si/25/made/en/ Micheletti, N., Chandler, J. H. and Lane, S. N. (2015) ‘Structure from Motion (SfM) Photogrammetry’, British Society for Geomorphology Geomorphological Techniques, 2(2), pp. 1–12. doi: 10.5194/isprsarchives-XL-5-W4-37-2015. Bláha, M. et al. (2012) ‘Direct Georeferencing of Uavs’, ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXXVIII1/(September), pp. 131–136. doi: 10.5194/isprsarchives-XXXVIII-1-C22-131-2011. Royal Institution of Chartered Surveyors (2014) ‘Measured surveys of land, buildings and utilities - 3rd edition’, Measured surveys of land, buildings and utilities - 3rd edition. Available at: http://www.rics.org/uk/knowledge/professional-guidance/guidancenotes/measuredsurveys-of-land-buildings-and-utilities-3rd-edition/. Trimble (2018) Direct Georeferencing for UAV’s. Available at: https://www.applanix.com/products/dg-uavs.htm (Accessed: 19 April 2018).
20 Address Offices in Brussels : Rue du Nord 76, BE – 1000 Bruxelles. Tel +32/2/217.39.72 Fax +32/2/219.31.47 E-mail: maurice.barbieri@clge.eu - www.clge.eu EU-Transparency Register of interest representatives - 510083513941-24