Longitudinal Motion Control of ARVs Via Sliding Mode Controller Driven By Nonlinear Disturbance Observer

Underwater vehicles can be divided into the autonomous underwater vehicles (AUVs) and the remotely operated vehicles (ROVs). The main feature of AUVs is that it can achieve autonomous motion, but it cannot communicate with the mother ship in real time. So, the recorded information can only be obtained after it is docked. The ROVs are equipped with umbilical cables, it can be operated manually in real time so they are often used for complex underwater operations. Nowadays, a new type of underwater vehicles named autonomous remotely vehicles (ARVs) have appeared. It has two operating modes, one is to connect the umbilical cable for manual control, and the other is to detach the umbilical cable and use its own battery to power it for autonomous motion. Therefore, design a control strategy with the objective of auto-operation is a key feature to increase ARVs’ autonomy and reliability(Zereik et al., 2018)

In practice, there are a number of technical challenges in the control of underwater vehicles, such as the unknown external disturbances and system uncertainties. External disturbances include the effects of environmental loads such as ocean currents and the disturbances generated by the umbilical cable. The system uncertainties are usually caused by the inaccuracy of hydrodynamic coefficients. At the same time, the ARVs have the similar structure to ROVs, so it suffered from the problem of highly coupled and nonlinear dynamics. The above problems all lead to considerable difficulties in ARVs’ automatic control. To solve these problems, many advanced control which have been applied to underwater vehicles control can also be used in ARVs, such as adaptive control(Bessa et al., 2008, 2010; Chen et al., 2016; Chu et al., 2017), sliding mode control (SMC)(Soylu et al., 2008; Huang and Yang, 2018), fuzzy logic control(guo et al., 2003; Huo et al., 2018), backstepping control(Liu et al., 2016; Wang et al., 2015), and MPC control(Shen et al., 2018). SMC approach for underwater robot control has been widely used due to its simple form, strong robustness, competence for systems with profound nonlinearity and modeling uncertainty. These characteristics are considered very suitable for ARV motion control. However, we have to take careful considerations to avoid the chattering problem when utilizing SMC based method, because it will course high energy consumption, increased friction and thruster damage. The source of chattering generated by SMC is that the switching term contained in the controller is formed by a discontinuous function (such as the symbol function sign). This bang-bang manner not only guaranteed robustness but also caused actuators chattering. Therefore, there are two basic methods to deal with the chattering problem: boundary layer method and continuous switching term method (Slotine et al., 2004). Both methods are proposed that approximated the discontinuous sign function by smooth terms however at the price of deterioration in performances. In order to avoid chattering while mitigate the drawbacks of performance deterioration, many novel control strategy have been proposed. Among them, a double-loop sliding mode control method is widely used in the motion control of underwater vehicles. The principle of this method is to decompose the motion system into kinematics and dynamics loops, and design controllers for the two loops respectively. External disturbances and system uncertainties are eliminated in the inner loop, which can guarantee the higher tracking accuracy in outer loop(Qiao and Zhang, 2018). Huang and Yang (2019) proposed a double-loop sliding mode controller for the ROV to achieve trajectory tracking. A novel continuous switching function is utilized to replace the discontinuous terms. In addition to improving switching term performance, this strategy can effectively eliminate chattering phenomenon. However, this method requires the knowledge of upper bound of the total disturbance. Yang et al.,(2016) proposed a nonlinear robust controller for solving the longitudinal motion control problem of AUV. It adopted the double-loop sliding method as well as combine theory of high order sliding mode with the recurrent hermite neural network. This method can remove chattering and improve convergence rate but has a lot of parameter to be set, which will affect its performance in actual experiments. Yan et al., (2019) proposed an adaptive double-loop integral sliding mode control for AUVs. A novel direct adaptive neural network controller combined with a conditional integrator is presented, which provides the robustness and adaptation for the vehicle. The disadvantage of this method is that the design process is complicated.