Many potentially useful applications that broadly fall under the umbrella of payload transportation operations have been proposed for aerial towed-cable (ATC) systems, namely the precise capture and delivery of payloads. There remain outstanding issues concerning the dynamics and control of ATC systems that are inhibiting the near-term demonstration of these applications. The development of simplified representations of ATC systems that retain the important dynamics, yet are simple enough for use in control system development is limited. Likewise, little research exists into the development of controllers for ATC systems, especially the development of towing strategies and cable-based control techniques for rendezvous and payload transportation. Thus, this thesis presents novel research into the development of control strategies and simulation facilities that redress these two major anomalies, thereby overcoming a number of hitherto unresolved issues. The primary objective of this thesis is to develop innovative non-linear optimal control systems to manoeuvre a cable towed beneath an aircraft to transport payloads both to and from surface locations. To appropriately satisfy this objective, accurate and efficient modelling capabilities are proposed, yielding the equations of motion for numerous models of the ATC system. A series of techniques for improving the representativeness of simple dynamical models were developed. The benefits of using these procedures were shown to be significant and possible without undue complexity or computational expense. Use of such techniques result in accurate simulations and allow representative control systems to be designed. A series of single and multi-phase non-linear optimal control problems for ATC systems are then formally proposed, which were converted into non-linear programming problems using direct transcription for expedient solution. The possibility of achieving accurate, numerous instantaneous rendezvous of the cable tip with desired surface locations on the ground, in two and three-dimensions, is successfully demonstrated. This was achieved through the use of deployment and retrieval control of the cable and/or aircraft manoeuvring. The capability of the system to safely and accurately transport payloads to and from the surface via control of the cable and/or aircraft manoeuvring is also established. A series of parametric studies were conducted to establish the impact that various parameters have on the ability of the system to perform various rendezvous and payload transportation operations. This allowed important insights into to the nature of the system to be examined. In order for the system to perform rendezvous and payload transportation operations in the presence of wind gusts, a number of simple closed loop optimal feedback controllers were developed. These feedback controllers are based on the linear quadratic regulator control methodology. A preliminary indication of the robustness of ATC systems to wind gusts is provided for through a succession of parametric investigations. The performance of the closed-loop system demonstrates that precise and robust control of the ATC system can be achieved for a wide variety of operating conditions. The research presented in this thesis will provide a solid foundation for further advancing the development of aerial tether payload transportation technology.

We propose a Model-Based Deep Reinforcement Learning (MBDRL) framework for collaborative paylaod transportation using Unmanned Aerial Vehicles (UAVs) in Search and Rescue (SAR) missions, enabling heavier payload conveyance while maintaining vehicle agility. Our approach extends the single-drone application to a novel multi-drone one, using the Probabilistic Ensembles with Trajectory Sampling (PETS) algorithm to model the unknown stochastic system dynamics and uncertainty. We use the Multi-Agent Reinforcement Learning (MARL) framework via a centralized controller in a leader-follower configuration. The agents utilize the approximated transition function in a Model Predictive Controller (MPC) configured to maximize the reward function for waypoint navigation, while a position-based formation controller ensures stable flights of these physically linked UAVs. We also developed an Unreal Engine (UE) simulation connected to an offboard planner and controller via a Robot Operating System (ROS) framework that is transferable to real robots. This work achieves stable waypoint navigation in a stochastic environment with a sample efficiency following that seen in single UAV work. This work has been funded by the National Science Foundation (NSF) under Award No. 2046770. / Master of Science / We apply the Model-Based Deep Reinforcement Learning (MBDRL) framework to the novel application of a UAV team transporting a suspended payload during Search and Rescue missions. Collaborating UAVs can transport heavier payloads while staying agile, reducing the need for human involvement. We use the Probabilistic Ensemble with Trajectory Sampling (PETS) algorithm to model uncertainties and build on the previously used single UAVpayload system. By utilizing the Multi-Agent Reinforcement Learning (MARL) framework via a centralized controller, our UAVs learn to transport the payload to a desired position while maintaining stable flight through effective cooperation. We also develop a simulation in Unreal Engine (UE) connected to a controller using a Robot Operating System (ROS) architecture, which can be transferred to real robots. Our method achieves stable navigation in unpredictable environments while maintaining the sample efficiency observed in single UAV scenarios.