Control Architecture and Path Planning for Quadcopters in Formation

Abstract

Unmanned aerial vehicles (UAVs) have found many areas of operation with numerous studies available in the literature. However, increasing demands in applications and the rapid development of technologies have transcended the use of a single UAV to the formation and their coordination. In the literature, UAVs’ low-level control, path planning, and formation maintenance have been addressed mainly in separation. This research proposes a control architecture that integrates those three subsystems with a task assessment unit and communication links to accommodate a variety of applications.

At the low level, robustness of the UAV control systems is important for applications which require accurate attitudes, also for safety maintenance and configuration preservation when flying in formation. In operations, UAVs are often subject to nonlinearity, external disturbances, parametric uncertainties and strong coupling, which may downgrade their control performance. Therefore, the first focus is to design robust control schemes to track desired attitudes under various conditions. Accordingly, robust low-level controllers for UAVs are developed, namely the adaptive quasi-continuous and adaptive twisting sliding mode control. They offer a novel technique to adaptively change the control parameters of the so-called sliding modes for the sake of performance improvements

To deploy multiple-UAV systems, the proposed control architecture includes robust control, path planning, and formation maintenance to create a real-time system that can be used for many engineering purposes. The system coordinates multiple UAVs in a specific formation to collect data of the inspected objects. The hardware extension on the basis of 3DR Solo drones includes the Internet of Things (IoT) and environmental sensors. Communication links are implemented by employing IoT boards for components of the control architecture to equip them with network and data processing capabilities.

For UAV formation control, a novel multi-objective angle-encoded particle swarm optimisation algorithm is proposed to generate formation trajectories. Here, the algorithm is developed to minimise a cost function incorporating multiple objectives subject to formation constraints that include inspection task completion, shortest paths and safe operation of the drones.

To handle difficulties arising from various inspection surfaces, avoid possible dynamic collisions, and maintain safe motion of the whole UAV formation, the path planning algorithm is incorporated with a reconfigurable capability developed to be integrated to the control architecture. This integration allows for flexible changing of the formation to accommodate additional constraints on collision avoidance, flight altitude, communication range, and visual inspection requirements.

Throughout the dissertation, analytical work developed is validated by extensive simulation, comparisons and experiments to evaluate the proposed approach and confirm its feasibility and effectiveness. Discussions on theoretical aspects and implementation details are included together with some recommendations.