Autonomous Aerial Vehicles

Research in autonomous aerial vehicles has grown rapidly and dramatically the past decade, aiming to address several challenges and problems where endurance is needed beyond endurance limit of human body or where there is not any necessity of human presence. Todays, UAVs play undebatable role in human life because of their utilizing in essential and commercial applications such as monitoring (i.e. traffic, shipping emission, oil spill), inspection (i.e. wind turbine,live gas flare), agriculture, search and rescue operations(i.e. finding victims, delivering help, first aid), building exploration, Reforestation, Cinematography, mapping and so on [1].

Conventional unmanned vertical take-off and landing vehicles can be classified according to their number of actuators into one of four categories: ducted Fan UAV (equipped with one propeller) [2-4], conventional helicopter (equipped with two propellers)[5, 6] , tricopter (equipped with three propellers ) [7, 8], quadrotors (equipped with four propellers)[9, 10]. The mentioned four categories of UAVs are under-actuated and therefore, they have some limitations on their performance because of this fact that they actuated with the number of actuators fewer than six DOF in space (i.e. they couldn’t be able to keep roll and pitch at zero angle while making translational motion along an arbitrary axis (i.e. x, y, z)). Consequently, it is unrealistic to change the orientation of under-actuated flying robot to accomplish difficult grasping missions while translation motion [11].

In order to overcome complex manipulation tasks, an over-actuated Micro Aerial Vehicle called the omnicopter based on novel mechanical design has been first developed by Long et al. [12]. In the proposed design, it is used five rotors, adjusting the yaw angle and also making trust to hover are provided through locating two fixed major coaxial counter-rotating propellers in the center of flying robot. Control of the roll and pitch and providing lateral forces are also accomplished through locating three adjustable angle ducted fans in three places surrounding the airframe of flying robot. Comprehensive details of description and design of the presented over-actuated flying robot can be seen in [11-17]. It should be pointed out that configuration of three ducted fans is considered such that symmetry satisfied.

As described before, vertical take-off and landing vehicles have many applications but successful utilizing of them in these applications require good flight control, both position and attitude control capabilities. Although some researchers strived to overcome this problem using linear control methods, owing to its straightforward design and implementation procedures. However, these simplifications limit the application scope of the proposed control strategies. On the other hand, the difficulty of the controller design increases due to the un-modeled dynamic nonlinearity, parametric uncertainty and external disturbances. This problem has received special attention from flight control researchers and engineers for a special type of flying robot, named quadrotor or quadcopter.

Gabe Hoffmann et al. [18] controlled the attitude and altitude of X4 flyer using sliding mode control (SMC) and LQR control strategies, respectively. Samir Bouabdallah et al. [10] used two different control techniques, backstepping and sliding-mode control methods,  to control the position and attitude of a micro quadrotor called OS4 in the presence of disturbance. The effectiveness of proposed controllers was revealed successfully in both theoretical and experimental. Adigbli et al. [19], used sliding mode and backstepping control strategies in order to control the attitude and position of a micro quadrotor, considering external disturbance. The designed controllers were successfully implemented in practical. More literature on employment of other control structures for quadrotors can be found in [20]. Many researchers selected Euler angles to represent the attitude of flying robot and ignored the rotational kinematics, therefore a simplified mathematical model used in order to design a controller. A great work has been done by Yang and Yan [21], in which, the full dynamics of quadrotor with rotational kinematics were considered to design the controller. In their study, the attitude regulation problem of unmanned quadrotors is addressed through a novel adaptive fuzzy gain-scheduling sliding mode control strategy, considering inertial uncertainties and external disturbances. It should be noted that in their work the fuzzy logic combined with sliding mode control due to termination of the chattering phenomena in control inputs. Furthermore, it is clearly demonstrated via simulation results that the proposed is effective and robust against external disturbances.

In comparison with quadrotor, a little research works have been done on controlling of over-actuated flying robot. The main of them are as follows: Long et al. [12], presented a nonlinear model of over-actuated flying robot in state space using the quaternions and angular velocities as state variables and used feedback linearization method in order to attitude control of the robot. In another work, Long et al. [13], designed three different linear control schemes, PD control, Lyapunov-based control and optimal LQ control, based on a linearized mathematical model of the Omnicopter, with a fixed vertical ducted fan angle configuration. As the result, the effectiveness of designed controllers demonstrated through simulation results. Long and Cappelleri [14] presented mathematical nonlinear model of the Omnicopter in details and used backstepping based attitude controller and a standard PID position controller in order to full control of the robot. The path tracking showed through simulation results and the performance of designed attitude controller demonstrated experimentally. Long and Cappelleri [15] provided an improvement in dynamic modeling of over-actuated flying robots, taking into account the aerodynamic drag effects due to the ducts and the gyroscopic effects due to the tilting of the surrounding fans. Furthermore, they proposed a control allocation algorithm for the omnicopter under both fixed and variable ducted fan angels’ configurations, and implemented the proposed algorithm for the former configuration. Long and Cappelleri [16] presented a comprehensive description of Omnicopter as well as comparisons between different Micro Aerial Vehicles. Furthermore, the classical PID control method used to control design and allocation technique utilized in order to create a mapping between the actuator inputs and virtual control inputs. Finally, the simulation and experimental results revealed the performance of designed Omnicopter. Long et al. [11] designed two adaptive backstepping controllers for both position and attitude sub dynamics of Omincopter and used control allocation technique based on weighted pseudo-inverse matrix method to obtain actuator inputs according to obtained virtual control inputs. The stability of closed-loop system guaranteed using Lyapunov theorem and the effectiveness of proposed controllers showed by means of simulation results.

From control point of view, the major drawback of control design using classical techniques (linear and nonlinear techniques except sliding mode control) is that they cannot be able to guarantee the robustness of closed-loop system in presence of external disturbance and parameter uncertainty. On the other hand, robustness of closed-loop system is required due to using of flying robot for outdoor application. Consequently, in this paper, sliding mode control approach is used as the position and attitude stabilizers of Omnicopter. However, the chattering phenomena of the sliding mode control will reduce the accuracy and excite un-modelled dynamics. Therefore, a fuzzy logic control approach is used to schedule the control gains of switching function adaptively based on the sliding surfaces information and elimination of chattering phenomenon as well as providing a good dynamic response. To providing a mapping between real control inputs and virtual control inputs a control allocation is employed based on the reflective newton algorithm. Finally, it is illustrated through simulation results that the proposed control approach are effective in tracking predefined trajectory and robust against external disturbance and internal uncertainty.

Therefore, the major contributions of this study can be described as follows:

1. Development of the full kinematics and  dynamics equations of motion of  Omnicopter
2. Development two robust control schemes, sliding mode and adaptive fuzzy sliding mode controllers, to full control of  Omnicopter.
3. Design of control allocation method based on the reflective newton algorithm method to construct a mapping between the real and virtual control inputs