Reactive-Based Position Control of an Underactuated Quadrotor

Abstract

This paper addresses the problem of position control or waypoint tracking for an underactuated quadrotor. The proposed control law is what is known as reaction based in the way that the attitude system reacts to errors in the translational motion. This methodology requires no generation of desired attitude or angular velocity and the resulting control law is model-independent in that it does not require the knowledge of the inertia matrix. In addition the controller has a very simple structure making it suitable for small quadrotor platforms. Simulation results are provided and discussed to demonstrate the proposed method.

Path planning and reactive based control for a quadrotor with a suspended load

“…where y e = y − y d is the position error in the y n axis, and f n z as in (9). Since f n y has the same structure as f n x , the stability proof follows along the same lines as above, to conclude that the origin of the closed loop system in the plane yz n is asymptotically stable.…”

“…Hence, we divide the control law into tasks: 1) Compute a force that drives the translational load error to zero, reduces the load oscillations, and guarantee asymptotically stability. 2) With the total force described in the first task, control the quadrotor movement with a control law similar to [9].…”

“…In this work we follow the approach set up in [9], where the attitude control for a quadrotor is proposed as…”

“…Thus we aim for a powerful control strategy that also is simple to implement, where a suitable approach is reactive control. Reactive-based control has been used to control quadrotors previously; for instance, [9] presents a reaction-based control where the attitude subsystem reacts to errors in the translational motion. The strength in this methodology lies in no requirement of generating the desired attitude or angular velocity.…”

“…Also, in [10], a second-order sliding mode attitude controller is derived which exponentially stabilizes the attitude at hover, and a PD controller for translational error is used to inject angular accelerations into the rotational dynamics. In this paper, we follow the ideas developed in [9], which computes the total necessary force to move and reduce the load's oscillations without controlling the swing angles but inject damping in the angular swinging velocity. More precisely, swing attenuation is performed in the plane projections (xz n and yz n ) of the swing angles, setting a non-restrictive control over the swing angles.…”