Fontanella, Rita (2018) Innovative Solutions for Navigation and Mission Management of Unmanned Aircraft Systems. [Tesi di dottorato]

Preview

Text Fontanella_Rita_31.pdf Download (8MB) | Preview

Collection description

The last decades have witnessed a significant increase in Unmanned Aircraft Systems (UAS) of all shapes and sizes. UAS are finding many new applications in supporting several human activities, offering solutions to many dirty, dull, and dangerous missions, carried out by military and civilian users. However, limited access to the airspace is the principal barrier to the realization of the full potential that can be derived from UAS capabilities. The aim of this thesis is to support the safe integration of UAS operations, taking into account both the user's requirements and flight regulations. The main technical and operational issues, considered among the principal inhibitors to the integration and wide-spread acceptance of UAS, are identified and two solutions for safe UAS operations are proposed: A. Improving navigation performance of UAS by exploiting low-cost sensors. To enhance the performance of the low-cost and light-weight integrated navigation system based on Global Navigation Satellite System (GNSS) and Micro Electro-Mechanical Systems (MEMS) inertial sensors, an efficient calibration method for MEMS inertial sensors is required. Two solutions are proposed: 1) The innovative Thermal Compensated Zero Velocity Update (TCZUPT) filter, which embeds the compensation of thermal effect on bias in the filter itself and uses Back-Propagation Neural Networks to build the calibration function. Experimental results show that the TCZUPT filter is faster than the traditional ZUPT filter in mapping significant bias variations and presents better performance in the overall testing period. Moreover, no calibration pre-processing stage is required to keep measurement drift under control, improving the accuracy, reliability, and maintainability of the processing software; 2) A redundant configuration of consumer grade inertial sensors to obtain a self-calibration of typical inertial sensors biases. The result is a significant reduction of uncertainty in attitude determination. In conclusion, both methods improve dead-reckoning performance for handling intermittent GNSS coverage. B. Proposing novel solutions for mission management to support the Unmanned Traffic Management (UTM) system in monitoring and coordinating the operations of a large number of UAS. Two solutions are proposed: 1) A trajectory prediction tool for small UAS, based on Learning Vector Quantization (LVQ) Neural Networks. By exploiting flight data collected when the UAS executes a pre-assigned flight path, the tool is able to predict the time taken to fly generic trajectory elements. Moreover, being self-adaptive in constructing a mathematical model, LVQ Neural Networks allow creating different models for the different UAS types in several environmental conditions; 2) A software tool aimed at supporting standardized procedures for decision-making process to identify UAS/payload configurations suitable for any type of mission that can be authorized standing flight regulations. The proposed methods improve the management and safe operation of large-scale UAS missions, speeding up the flight authorization process by the UTM system and supporting the increasing level of autonomy in UAS operations.

Downloads

Actions (login required)

View Item