Multi-Task Regression-based Learning for Autonomous Unmanned Aerial Vehicle Flight Control within Unstructured Outdoor Environments

Increased growth in the global Unmanned Aerial Vehicles (UAV) (drone)\nindustry has expanded possibilities for fully autonomous UAV applications. A\nparticular application which has in part motivated this research is the use of\nUAV in wide area search and surveillance operations in unstructured outdoor\nenvironments. The critical issue with such environments is the lack of\nstructured features that could aid in autonomous flight, such as road lines or\npaths. In this paper, we propose an End-to-End Multi-Task Regression-based\nLearning approach capable of defining flight commands for navigation and\nexploration under the forest canopy, regardless of the presence of trails or\nadditional sensors (i.e. GPS). Training and testing are performed using a\nsoftware in the loop pipeline which allows for a detailed evaluation against\nstate-of-the-art pose estimation techniques. Our extensive experiments\ndemonstrate that our approach excels in performing dense exploration within the\nrequired search perimeter, is capable of covering wider search regions,\ngeneralises to previously unseen and unexplored environments and outperforms\ncontemporary state-of-the-art techniques.\n