| Summary: | Small unmanned air systems (UAS), due to their navigational versatility and ability to operate autonomously, serve as an intriguing platform on which to carry out advanced sensing operations in otherwise untraversable or prohibitively dangerous environments. The need for UAS to be able to autonomously navigate and explore their environments with limited payload, communication, and computational capacity, however, poses its own challenges–particularly when subjected to the non-ideal environmental disturbances and feature spaces present in real-world scenarios. This thesis addresses these issues by presenting two complementary projects enabling UAS-based autonomous sensing in real-world environments using relatively low-cost and lightweight hardware. The first project presents a UAS capable of measuring air wakes while flying tethered behind a moving vessel. The unique challenges of tethered flight control and relative state estimation in a feature-starved environment are addressed with a novel planning and control architecture together with an error-state Kalman filter that achieves centimeter-level relative position accuracy. The second project presents a multi-agent UAS navigation system for GPS-denied environments that expands on the state-of-the-art in collaborative simultaneous localization and mapping (CSLAM) for the purpose of facilitating fast and accurate radiation mapping in contaminated and cluttered zones. CSLAM capabilities are made more robust to communication deficiencies through the novel incorporation of ultra-wideband range sensors into a distributed range-enhanced pose graph optimization (DRPGO) scheme. The experimental demonstrations of the two presented systems, considered in tandem to overcome hurdles to sensing from aerodynamic disturbances, feature-starved environments, and communication bandwidth limitations, strengthen the promise of small UAS as an effective tool for demanding real-world data collection applications.
|
|---|