Hexacopter outdoor flight test results of an Extended State Observer based controller

Abstract

Extended State Observers (ESO) provide the appealing capability of estimating states and disturbances (like modeling errors) simultaneously. In this paper we present a modified fullstate linear ESO as an augmentation of a baseline attitude tracking controller for a hexacopter. The baseline controller is designed using Nonlinear Dynamic Inversion with relative degree 2. By feeding back the estimated modeling error from the ESO, the performance as well as the robustness of the overall control system is significantly improved. These capabilities are demonstrated in outdoor flight tests with a hexacopter, where controlled flight is achieved even under severe actuator degradation. Due to the linear structure of the presented ESO, this might be an alternative to adaptive controllers in view of certification issues.

Fault-tolerant position tracking of a hexacopter using an Extended State Observer

“…In this paper, an extension of the attitude controller presented in [22] to a position tracking controller is presented. In this approach the baseline controller is augmented by a modified LESO, which increases the robustness without resorting to reconfiguration or high-gain feedback.…”

“…In the next two sections, the earlier results from the authors [22] will be reviewed for the sake of completeness. The baseline controller is based on Nonlinear Dynamic Inversion for the output y = η (see e.g.…”

“…It commands the desired attitude to the inner loop using a nonlinear mapping. The inner loop is a NDI with relative degree 2, which is augmented by a Linear Extended State Observer (LESO) as previously presented by the authors [22].…”

A novel extended state observer

“…In practice, the conventional PID controllers often provide satisfactory performance and robustness in the presence of small modeling error. However, in case of large modeling uncertainties and external disturbances, these simple controllers do not provide satisfactory disturbance rejection capability [3]. As pointed out in [4], many uncertainties and disturbances in control systems are unmeasurable and the disturbance estimation technique is particularly crucial for disturbance attenuation.…”

Adaptive Fault–Tolerant Position Control of a Hexacopter Subject to an Unknown Motor Failure

“…Most of them are partial results showing only failure detection and identification (Freddi et al, 2014), or only controller reconfiguration assuming that the fault has been already identified (Achtelik et al, 2012;Du et al, 2015;Schneider et al, 2012;Santos et al, 2015), or using external sensors for having an accurate position measurement (Amoozgar et al, 2012;Dydek et al, 2010;Mueller and D'Andrea, 2014;Saied et al, 2015;Vey and Lunze, 2016), or making computations off-board (Dydek et al, 2013). Furthermore, in the special case of the hexarotor, the controllers are applied to different levels of actuator faults: only degradation (Heise et al, 2014;Mühlegg et al, 2015), the total failure of one rotor excluding a group of rotors (Vey and Lunze, 2016;Yang et al, 2016), the total failure of any rotor (Falconí et al, 2016) and the total failure of more than one rotor (Mueller and D'Andrea, 2014). The complexity increases with each of these cases because after a rotor failure the set of attainable forces and moments is noticeably reduced and saturations become an issue.…”