Autonomous and high-precision mapping via vision-guided unmanned aerial systems

点击翻译按钮获取中文摘要

In recent years, integrating three-dimensional (3D) vision to unmanned aerial vehicles (UAVs) has contributed a great deal to the advancement of geo-spatial technologies for fine-scale mapping. Despite the recent efforts, exploiting these systems to their maximum potential has remained limited due to the lack of extensive research and development with respect to the following aspects: increasing the precision, safety and intelligence of navigation; improving the quality and autonomy of data acquisition; integrating multiple complementary data-collection technologies; enhancing techniques of data processing to raise the level of accuracy, completeness, and interpretability of the outputs. In this regard, the long-term objective of my research program (10+ years) is to develop novel approaches for emerging cognitive and collaborative 3D-vision solutions in geomatics engineering. The geomatics applications of these solutions include the broad areas of infrastructure inspection, law enforcement, search and rescue, on-demand emergency mapping, city modeling, wildlife management, and precision farming. To meet my long-term research goals, this short-term research program will address the challenges related to the development of intelligent stereo-vision systems based on unmanned aerial vehicles. Particularly, the thematic application of the proposed solutions will be high-precision surveying and metric inspection of urban infrastructure. Examples of such structures are vertically extended ones (e.g. telecommunication towers and building facades) and linearly extended ones (e.g. power lines and roads). The specific objectives of this research program include the following: 1) Developing vision-based techniques of precise pose estimation and navigation, 2) Task-based active view planning for autonomous image acquisition, and 3) Semantic stereo-vision computation for automated and accurate 3D mapping of infrastructure. The outcomes of the proposed research have the potential to significantly advance and revolutionize the techniques of infrastructure mapping and inspection. The systems and approaches developed in this research program will result in the autonomous production of high-accuracy 3D measurements and collection of detailed visual data from infrastructure. The solutions can easily compete with the current human-centric methods of observation and will eliminate the need to put lives of inspectors at risk to manually inspect hard-to-reach assets. In addition to infrastructure industries, the achievements of this research program are transferable to other domains of engineering wherever there is a need for high-precision autonomous mapping. The HQP trained in this program will be able to transfer their knowledge and skills to Canadian industries in the vast fields of applications for mapping and monitoring via modern geo-spatial technologies.