Nonlinear Fault-Tolerant Control of a Quadrotor UAV Based on Sliding Mode Control Technique Tong Li*, Youmin Zhang*, and Brandon W. Gordon* * Concordia University, Montreal, Quebec H3G 1M8, Canada (e-mail: ymzhang@encs.concordia.ca) Abstract: In this paper, a sliding mode based fault-tolerant control (SM-FTC) is designed, implemented and tested in a quadrotor unmanned small helicopter under the propeller damage and actuator fault conditions. Based on the concept of sliding mode control, both passive and active fault-tolerant control laws have been designed and experimentally testbed on a quadrotor UAV (unmanned aerial vehicle) testbed available at Concordia University. These two types of controllers are carried out and compared through both theoretical analyses and experimental tests on the quadrotor UAV system. 1. INTRODUCTION Early research has shown some design techniques for the fault-tolerant control system (FTCS). A recent comprehensive survey on FTCS is presented in (Zhang and Jiang, 2008) which classifies fault-tolerant control (FTC) as passive fault-tolerant control (PFTC), and active/ reconfigurable fault-tolerant control (AFTC) with a fault detection and diagnosis (FDD) scheme in a general framework. Safety, reliability and reconfigurability analyses are also outlined in the paper to make a link for the currently individual research work between control engineering and safety engineering. Some key points in FTCS were also summarized in an early review paper (Patton, 1997) for summarizing control design methods developed up to 1997. Zhang (2010) outlined a general fault modeling method in FTCS for three different situations on sensor faults, actuator faults, and system dynamic faults. Fekih and Pilla (2007) presented a passive fault-tolerant control methodology using sliding surface and Lyapunov function to eliminate the pre- specified faults for an F-18 aircraft model. The results show the effectiveness of the control design. Zhang and Jiang (2002) presented an integrated design procedure for fault detection, diagnosis, and reconfigurable control. A two-stage adaptive Kalman filter is used in fault detection and diagnosis scheme. The reconfigurable feedback and feedforward controllers are also developed in details. Milhim et al. (2010) designed a gain scheduling based PID (GS-PID) controller for FTC of a quadrotor UAV under simulation environment. The idea of GS-PID has been further implemented and experimentally tested in a quadrotor helicopter UAV in Sadeghzadeh et al. (2011) with three-dimensional trajectory tracking capability in the presence of actuator faults. In Alwi and Edwards (2005), a sliding mode based fault-tolerant control has been designed for a civilian fixed-wing aircraft, Boeing 747. The elevator failure is simulated and the simulation results show that the performance of the controller is good. Alwi and Edwards (2008) proposed another method using sliding mode scheme with control allocation for FTC of the Boeing 747. With on-line control allocation, an active fault-tolerant control has been successfully designed and illustrated using sliding mode algorithm. The safety and reliability is and will always be a critical issue in the aviation industry. Therefore, in this paper, in order to further demonstrate the capability of fault-tolerant control systems on handling faults in aviation systems, a sliding mode control based PFTC and AFTC strategies are designed, implemented and tested in a quadrotor helicopter UAV testbed available at Concordia University, respectively. The quadrotor UAV used in the paper is also known as Qball-X4. All the designs and experiments have been carried out in this physical Qball-X4 system. 2. MODEL OF THE QUADROTOR UAV The groundwork of a controller designing process is always based on a mathematical model of the system to be controlled. In this paper, a dynamic model is needed because forces generated by four propellers are the main reasons that the quadrotor flies and these propellers need to be controlled in appropriate ways for different flight modes and flight conditions. Fig.1 shows more clearly on the relation between movements and forces. Positive direction of pitch, roll and yaw angles have been presented as marked in the figure. Fig. 1. Definition of the Qball-X4 attitude Qball-X4 is a rigid body, and two sets of reference frames have been used to formulate the system dynamic equations. One frame is the body-fixed frame in which the origin is located at the centre of the mass of the Qball-X4, as shown in Fig. 1. The other reference frame is the earth-fixed frame (or known as global frame) in which the origin can be chosen as desired. The coordinates , , q q q x y z are defined in body-fixed 8th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes (SAFEPROCESS) August 29-31, 2012. Mexico City, Mexico 978-3-902823-09-0/12/$20.00 © 2012 IFAC 1317 10.3182/20120829-3-MX-2028.00056