Download files
Abstract
3D mapping using small UAV as small platforms carrying navigation and imaging sensors has gained the popularity in the recent years. However, the improvement in 3D UAV mapping to obtain more accurate and reliable results can increase the benefits gained from them leading the 3D UAV mapping suitable for more applications. The improvement can be introduced through many ways, but this research will try to deal with multiple gross errors that can occur in navigation or geo-referencing step and systematic and gross errors in mapping step to obtain robust results. The major research contributions are summarised as follows:
a. New analysis for three and four fault cases:
The comprehensive investigation of the characteristics of multiple faults in navigation sensors GPS, GNSS, GPS/INS, and GNSS/INS shows that the integration of INS with GNSS systems can make the identification of faulty measurements easier compared with the use of standalone GNSS system. The integration of INS can also improve the reliability in navigation system in the case of three and four faults.
b. A new factored stochastic model:
A new factored stochastic model has been proposed in this study to deal effectively with the random errors that can affect mapping process. It was based on the horizontal positions of the objects in the mapped area. However, more complex stochastic models may be further investigated. Furthermore, the impact of the accuracy of geo-referencing and matching observations has been discussed to give an indication about the future of mapping using UAVs.
c. A new procedure for testing the systematic error parameters:
A new comprehensive procedure for testing the significance of the systematic error parameters to choose the necessary or the most significant parameters to model the systematic errors has been introduced in this study. The procedure has been tested under different cases of the distribution of GCPs. It also has been investigated with different scenarios of simulated systematic errors.
d. A comprehensive statistical quality analysis for 3D UAV mapping:
The comprehensive statistical quality analysis for 3D UAV mapping in the case of single and double fault cases has been completed showing that the estimation of 3D coordinates using only two overlapped images can provide less reliable or unreliable solution in single fault case. In the double fault case, the situation becomes worse in terms of the reliability. The analysis shows that the reliance on the increased number of overlapped images can be considered as the best practice to improve the reliability in single and multiple fault cases.