Input Allocation for the Propeller-Based Overactuated Platform ROSPO

Abstract

We apply input allocation to a redundantly actuated platform driven by tilting aerodynamic propulsion units: the ROtor graSPing Omnidirectional (ROSPO). This platform represents a novel testbed for redundancy allocation designs in propeller driven platforms. The control solution is based on a hierarchical architecture, made of a high level controller for trajectory tracking, and a nonlinear input allocation algorithm. The algorithm exploits the input redundancy to take into account soft constraints associated to physical saturation limits of the actuators, and also induce reduced energy consumption. The actuator dynamics is fully taken into account in the framework and a rigorous proof of asymptotic tracking of time-varying references is guaranteed despite the impossibility of an instantaneous force execution. The experiments on the ROSPO platform clearly show the practicability and effectiveness of the proposed approach, as well as its scalability with different degrees of over-actuation levels.

Dual-layer optimization-based control allocation for a fixed-wing UAV

“…Plant Actuators High-level Controller The challenges as mentioned above are, at first glance, similar to the problems studied in another branch of allocation, with the name of dynamic control allocation [4,7,16,17,18,19,20,21,22,23,24], which emphasizes that the actuator dynamics are taken into consideration in the allocation design. However, the main difference is that the 'actuator dynamics' in the Aerosonde aircraft is artificially built to ensure that the relative degrees of the altitude and velocity are well defined.…”

“…In [22], the output regulation problem for over-actuated linear systems is solved, which can be applied to the case of linear actuators but without constraints. In [23,24], the optimality-based dynamic allocation is designed for redundancy actuators with the first-order nonlinear dynamics, where the actuator constraints are involved via suitable penalty terms in the cost function. However, similarly to the pioneering work [16], offset-free allocation can only be achieved in the case of constant control signals [24, Th.…”

“…Remark 4. If the constraints in the outer-layer QP (23) are not active, only the inner-layer one (24) operates in the whole allocation process. In such a case, the proposed method will degenerate into the conventional control allocation method [12, Sec.…”

Comparison of Control Methods for Trajectory Tracking in Fully Actuated Unmanned Aerial Vehicles

“…By referring to the simplified dynamical model described by equations ( 11)-( 12), one sees that the attitude dynamics (12) evolves independently from the translational one (11). Moreover, since the control force f c can be delivered in any direction with respect to the airframe (thanks to full actuation) and R is measured, it is possible to use the virtual input f d := R f c as control variable for the position dynamics.…”

“…Clearly, combining a too aggressive controller and significant initial pose errors would require fast and large tilt-angle variations θ a , which could be impossible to track: the underlying tradeoff between performance and stability should be considered in the control design validation. Clearly, including the actuators dynamics (2)-(3) in the control design itself would be more accurate; however, it would also make the overall architecture more complex than needed, at least for standard applications [11]. Note, in passing, that motors and servo-actuators have similar bandwidth for the current application although servos have a more complex dynamical behavior than motors (see Section VI-A and [1]).…”