Design of passive fault-tolerant controllers of a quadrotor based on sliding mode theory

Abstract

In this paper, sliding mode control is used to develop two passive fault tolerant controllers for an AscTec Pelican UAV quadrotor. In the first approach, a regular sliding mode controller (SMC) augmented with an integrator uses the robustness property of variable structure control to tolerate partial actuator faults. The second approach is a cascaded sliding mode controller with an inner and outer SMC loops. In this configuration, faults are tolerated in the fast inner loop controlling the velocity system. Tuning the controllers to find the optimal values of the sliding mode controller gains is made using the ecological systems algorithm (ESA), a biologically inspired stochastic search algorithm based on the natural equilibrium of animal species. The controllers are tested using SIMULINK in the presence of two different types of actuator faults, partial loss of motor power affecting all the motors at once, and partial loss of motor speed. Results of the quadrotor following a continuous path demonstrated the effectiveness of the controllers, which are able to tolerate a significant number of actuator faults despite the lack of hardware redundancy in the quadrotor system. Tuning the controller using a faulty system improves further its ability to afford more severe faults. Simulation results show that passive schemes reserve their important role in fault tolerant control and are complementary to active techniques.

Quadrotor Fault Tolerant Incremental Sliding Mode Control driven by Sliding Mode Disturbance Observers

“…Many other SMC/SMDO designs in the literature also contain a preliminary feedback term using NDI to reduce the control efforts of SMC/SMDO. For example, adaptive fuzzy gain-scheduling SMC [12], first-order SMC using the equivalent control estimated from the nominal model [4,6,7,8,9,19], adaptive SMC [5], higher-order SMC [20,21,22], adaptive super-twisting SMC [11], modified super-twisting SMC using a higher-order sliding mode observer [13].…”

“…SMC/SMDO is able to observe and compensate for bounded perturbations, as shown in Eqs. (9,10). Even though the SMC/SMDO control input designed by Eq.…”

“…In Ref. [4,5,7,9,10,11,12,14], the control input vector is taken as u = [M c , T ] T . This choice is deficient because only the uncertainties of I v can be considered in the controller designs.…”

“…Being invariant (better than just robust) to matched uncertainties [2,3], Sliding Mode Control (SMC) is a promising candidate to fulfill this goal. A variety of SMC methods have been proposed for quadrotors to resist external disturbances and to cope with faults [4,5,6,7,8,9,10,11,12,13,14]. In spite of the varieties in SMC designs, for most SMC algorithms, the required control gains are positively correlated with uncertainty bounds (for first-order SMC), or the bounds of uncertainty derivatives (for higher-order SMC).…”

“…Regrading nonlinear control problems, this feedback term is commonly derived by dynamically inverting nonlinear algebraic equations, namely, by using Nonlinear Dynamic Inversion (NDI). Examples can be given for both first-order [4,5,6,7,8,9,12,19] and higher-order [11,13,20,21] sliding mode control methods. The other approach is incorporating the uncertainty estimations, for example by using Sliding Mode Disturbance Observers (SMDO), such that only the estimation errors need to be dealt with by SMC [8,13,19,22].…”