APPROXIMATE INVERSE DYNAMICS LINEARIZATION OF AN AERIAL MANIPULATOR SYSTEM

Abstract

Aerial manipulators draw increasing attention due to their ability to achieve both aerial mobility and physical interaction with the world. With these capabilities, they offer solutions for tasks in hard-to-reach environments and are employed in various fields, including military applications, infrastructure inspection, construction and transportation. However, controlling these systems presents a lot of challenges due to their highly nonlinear, coupled, and time-varying dynamics. The existing nonlinear control methods, such as feedback linearization, require a perfectly known and continually updated dynamic model, which is computationally expensive and not always feasible for fast, real-world applications. This paper introduces a novel application of approximate inverse dynamics linearization to a quadrotor unmanned aerial vehicle (UAV) equipped with a 2 degree-of-freedom (DOF) robotic manipulator designed for grasping and manipulating different payloads. Our approach uses offline-computed, fixed nominal dynamics for the model linearization, treating the residual nonlinearities and inaccuracies as bounded structured uncertainties. This formulation simplifies the control problem, allowing for the design of robust controllers that can guarantee the stability and performance despite the dynamic changes of the system. Through numerical simulations, we demonstrate that this method achieves accurate trajectory tracking and effectively captures the influence of the manipulator motion on the system dynamics within the structured uncertainty. A key finding is the significant reduction in computational load, with approximate method being, on average, 50 times faster than online calculation of the full dynamics, proving its viability for real-time control applications.