While using multirotor UAVs for transport of suspended payloads, there is a need for stability along the desired path, in addition to avoidance of any excessive payload oscillations, and a good level of precision in maintaining the desired path of the vehicle. However, due to the nonlinear and underactuated nature of the system, in addition to the presence of mismatched uncertainties, the development of a control system for this application poses an interesting research problem. This paper proposes a control architecture for a multirotor slung load system by integrating a Multi-Surface Sliding Mode Control, aided by a Radial Basis Function Neural Network, with a Deep Q-Network Reinforcement Learning agent. The former will be used to ensure asymptotic tracking stability, while the latter will be used to suppress payload oscillations. First, we will present the dynamics of a multirotor slung load system, represented here as a quadrotor with a single pendulum load suspended from it. We will then propose a control method in which a multi-surface sliding mode controller, based on an adaptive RBF Neural Network for trajectory tracking of the quadrotor, works in tandem with a Deep Q-Network Reinforcement Learning agent whose reward function aims to suppress the oscillations of the single pendulum slung load. Simulation results demonstrate the effectiveness and potential of the proposed approach in achieving precise and reliable control of multirotor slung load systems.

Multi-surface sliding mode control addresses the limitations of traditional sliding mode control by employing multiple sliding surfaces to handle uncertainties, disturbances, and nonlinearities. The design process involves developing sliding surfaces, designing switching logic, and deriving control laws for each surface. In this paper, first, a robust finite-time multi-surface sliding mode controller will be presented and its performance analyzed by applying it to a multirotor subjected to a suspended payload, modeled in the form of a single pendulum, itself defined as a spatial (3D) dynamic model. Next, an adaptive finite-time multi-surface sliding mode controller will be derived—adding a variable adaptive parameter to the existing sliding surfaces of the robust finite-time control—and applied to the same system. It will be shown that the adaptive controller, with an adaptive parameter that adjusts itself based on the present value of the multi-surface sliding mode parameter, creates an improved fast finite-time convergence by obtaining an optimal settling time and minimizing undershoot of the multirotor state vector. Empirical verification of the effectiveness of the adaptive control will be carried out by presenting the control performances against a step response. It is also shown that the control may be utilized to approximate external disturbances—represented by the pendulum—and that with the application of control, the vehicle’s motion may be stabilized and the payload swing suppressed. Lyapunov stability theory-based stability proofs for the controllers’ designs are developed, showing the asymptotic stability of the output and uniform boundedness of the errors in the system dynamics. It is verified that the multi-surface sliding mode control can account for system uncertainties—both matched and mismatched—in addition to changes in internal dynamics and disturbances to the system, where the single pendulum payload is representative of the changes in dynamics that may occur to the system. Numerical simulations and characteristics are presented to validate the performance of the controllers.