| dc.description.abstract | Unmanned aerial vehicles (UAVs) have been widely used in reconnaissance, remote sensing, disaster relief, and agricultural applications. Many applications require multiple UAVs to cooperate to complete the assigned tasks. Two such applications of interest involving multi-UAV collaboration, which are studied extensively in the existing literature are payload transportation and collective target tracking. In this thesis, we study the design of control laws in the context of these applications. Both the applications involve the guidance of the UAVs to a goal or target, while they mutually interact and fly in a formation. The payload transportation problem addresses the issue of limited lifting capabilities of UAV systems, and the need to aggregate them to provide sufficient thrust to lift the payload. The target tracking problem, however, adds robustness to the target monitoring and estimation task with a team of UAVs. The utilization of multiple UAVs in these applications adds redundancy, and thus robustness to UAV failures. Operating such complex systems requires the design and implementation of practical motion control laws, which in turn warrant the design of simple formulations. Such formulations make the coordination of scaled UAV formations possible, besides providing a framework for accommodating the aspect of their safety in the control design.
We employ a two-pronged approach in the control design for payload transportation and target tracking problems. In the first part of the thesis, we develop distance and bearing-based proportional derivative (PD) motion control laws meeting each identified problem objective in a modular framework. In the second part of the thesis, we address the constraints on the UAVs in a formation, in the context of obstacle and inter-UAV collision avoidance, and actuator saturation. Our multi-objective modular control approach makes it possible to modify the control laws based on the objectives of the formation, making our work applicable to a variety of problems. A pair of UAVs that carry a payload and settle on standoff circles about a virtual target, while maintaining their mutual distance, is considered as an example application to demonstrate our approach. As compared to a leader-follower role assignment in the existing literature, we do not assign any roles to the UAVs, which are assumed to be homogeneous. Additionally, we design the guidance strategy for UAVs to reach a target, independent of the payload, relying on the UAV-payload geometry to guide the attached payload to the target location. We simulate agricultural spraying operations in an annular swathe, and over a field using the lawnmower trajectory, with dynamic UAV and payload models, to illustrate how the modular control design is practically applicable. The modular design is augmented by our utilization of non-smooth control laws using the signum and saturation nonlinearities, which further simplifies the control formulations in the payload transportation application, making real-world implementation practical. Use of such non-smooth motion control laws for UAV formations is not prevalent in the existing work, and thus is a major contribution of the thesis. Further, the Lyapunov stability analyses of the multi-objective smooth and non-smooth control laws using results from non-smooth analysis, illustrate the advantages of our modular control approach in yielding analytical guarantees of kinematic stability.
An additional benefit of our control approach for multi-UAV applications is in providing a platform for integrating operational and safety constraints. The aspect of safety in collaborative UAV applications has not been properly addressed in the existing literature. Formations often require the UAVs to maintain a safe distance from each other, while collectively staying away from obstacles in the environment. To this end, we design actuator saturation constraints, and safety constraints for obstacle and inter-UAV collision avoidance. Control barrier functions (CBFs) are used to develop and enforce these constraints in a control-optimization problem involving a payload which is rigidly attached to two UAVs which settle on target standoff circles, while navigating an environment with obstacles. We extend this optimization problem further and add control Lyapunov function (CLF) constraints, to a problem involving four UAVs and a semi-flexible payload requiring safe shape-manipulation to navigate between obstacles. We also explore the utilization of barrier Lyapunov functions (BLFs) in target tracking problems, where a formation of UAVs tracks a moving target while maintaining inter-UAV separation within specified bounds. The barrier functions provide implementable constraints that are easily integrated with our modular multi-objective control architecture, to provide real-time solutions for collaborative UAV applications. This is extensively demonstrated using numerical and ROS-Gazebo simulations with dynamic UAV and payload models, using the proposed controllers and constraints. Thus, by incorporating these barrier functions in our analysis, our second major contribution is the safety-critical control design of multi-UAV systems.
To summarize, the thesis contributes by developing simple, scalable, and implementable motion control laws to guide UAV formations towards a target, while maintaining their individual and collective safety in the collaborative payload transportation and target tracking applications. | en_US |
