Serdar Soylu

Robust Control of Underwater Vehicles with Fault-Tolerant Infinity-Norm Thruster Force Allocation

There are to objectives to this paper. First, a chattering-free sliding mode controller is proposed for the trajectory control of remotely operated vehicles (ROVs). Secondly, a new approach for thruster force allocation is proposed that is based on minimizing the linfin norm. With regards to the former, a new adaptive term is developed that eliminates the high frequency control action inherent in a conventional sliding-mode controller, and also removes the need for a priori knowledge of upper bounds on uncertainties in the dynamic parameters of the ROV. With regards to the latter, it is demonstrated that the linfin norm optimization can be cast as a linear problem that affords easy incorporation of the thruster saturation limits. Using numerical simulations, it is shown that the proposed linfin thruster allocation is capable of meeting the adaptive sliding mode controllers demands in the presence of thruster failures and is therefore fault tolerant. Finally, a recurrent neural network is designed in order to obtain a real time solution rate to the thruster allocation problem.

Comprehensive underwater vehicle-manipulator system teleoperation

In this work, a novel comprehensive scheme for the coordinated control of remotely operated vehicle-manipulator systems (ROVMs) is proposed. In the proposed scheme, instead of commanding the motion of the vehicle and the manipulator separately, a human pilot commands only the manipulators end-effector motion using a parallel- architectured six-degree-of-freedom (6-DOF) joystick. The generated reference motion is then converted into a set of desired ROV and manipulator joint motion by means of using a redundancy resolution scheme that provides the means to utilize redundant degrees of freedom to accomplish secondary objectives. The redundancy resolver uses the Gradient Projection Method combined with a Mamdani-based fuzzy determination of the hierarchy of the secondary objectives. The controller relies on a unified dynamic model of the system. The quasi-Lagrange method is used to derive the equations of motion in terms of the ROV body-fixed frame. For the control problem, a sliding-mode based controller is used that contains an adaptive term for the estimation of the upper bound on the lumped uncertainty vector. The hardware-in-the-loop simulation studies illustrate that detailed subsea tasks can be completed with a small, low-cost ROVM system using the proposed ROVM operation scheme.

Development of a coordinated controller for underwater vehicle-manipulator systems

In this work, the motions of an underwater remotely operated vehicle (ROV) and a spatial manipulator are coordinated using a consolidated controller. The controller translates a single pilot command, the desired position and orientation of the end effector, into a coordinated set of ROV and manipulator joint motions that satisfy the pilot intent in addition to a series of secondary objectives. The controller relies on a unified dynamic model of the system. The quasi-Lagrange method is used to derive the equations of motion in terms of the ROV body-fixed frame. For the control problem, a novel sliding-mode based controller is proposed. The controller contains two layers of adaptivity. The first layer is for adjusting PID gains, whereas the second layer is for estimating the bound on a lumped uncertainty vector. The second level of adaptation is shown to relax the Lyapunov stability requirements leading to a more robust controller. To generate reference state values, a redundancy resolution technique is utilized that is based on the gradient projection method merged with a fuzzy determination of the hierarchy of the secondary objectives. The redundancy resolution method distributes the pilots end-effector command over the ROV and the manipulator in an optimal manner using the redundant degrees of freedom. The results illustrate that detailed subsea tasks can be completed with a small, low-cost ROVM system using the proposed unified control scheme.

A Fault-Tolerant Fuzzy-Logic Based Redundancy Resolution Method for Underwater Mobile Manipulators

In this work, a fault-tolerant redundancy resolution scheme is presented that allows a single 6-DOF command to be distributed over a small URVM system composed of an otherwise underactuated URV and serial manipulator. The URVM system admits an infinite number of joint-space solutions for each commanded end-effector state due to its inherent redundancy. The primary objective is realized using the right Moore-Penrose pseudoinverse solution. The secondary objectives are: avoiding manipulator joint limits, avoiding singularity and high joint velocity; keeping the end-effector in sight of the on-board camera minimizing the URV motion; and minimizing the drag-force resistance, or weathervaning. Each criterion is defined within the framework of the Gradient Projection Method. The hierarchy for the secondary tasks is established by a low-level artificial pilot that determines a weighting factor for each criterion based on if- then type fuzzy rules that reflect an expert human pilots knowledge. A Mamdani fuzzy inference system is used to interpret the fuzzy rules based on the sensory knowledge. The resulting weight schedule yields a self-motion (null-space motion) that emulates how a skilled operator would utilize the full capabilities of the URVM to achieve the secondary objectives. The proposed redundancy resolution scheme has a fault-tolerant property. When a joint failure occurs, the scheme automatically redistributes the end-effector velocity command taking into account the faulty joints. To demonstrate the efficacy of the proposed scheme, several numerical simulations are performed The results illustrate the validity of the proposed redundancy scheme.

Dynamics and control of tethered underwater-manipulator systems

In this work, the dynamics modelling strategy of a tethered underwater remotely operated vehicle (ROV) coupled with a and spatial manipulator have been studied. With regards to the cable dynamic modelling, it is considered to be a series of lumped point masses connected by linear, massless, visco-elastic springs. In addition, the model accounts for the tether bending and twisting effects. Regarding the manipulator dynamics, the Articulated-Body Algorithm is employed due to its computational efficiency. In order to control the ROV motion under disturbance forces and moments caused by the tether and the manipulator motion, a series of Model-based SISO sliding-mode controllers are implemented and the ABA is used to predict the dynamic coupling force expressions based on the feedback of ROV and the manipulator states. The control gains of the sliding-mode controllers are defined as a closed function of the articulated inertias of the ABA algorithm; leading to time varying gains as opposed to constant as in conventional methods. As a case study, a Saab-Seaeye FALCON™ ROV with a modified Hydrolek™ HLK 43000 manipulator is presented. Numerical simulations are performed to reveal the extent to which the tether dominates the Falcon-manipulator dynamics. It is shown that disturbance forces and moments created by tether motion must be actively compensated using while the ROV is held stationary during manipulator operation. It is also shown that the use of force sensors at the FALCON™s tether termination can dramatically improve the performance of the series of SISO sliding mode controllers.

Automation of CRD100 seafloor drill

The main focus of this paper is to present control and automation aspects of the remotely operated CRD100 seafloor drill. The CRD100 is a fourth generation, state-of-the-art robotic seafloor drill that is directly deployed onto the seabed. It is designed to drill, case, core, and collect geotechnical sampling data to a depth of 65m below the seabed at 3000m beneath the water surface. The system must perform a number of repetitive tasks such as retrieving drilling tools from a rack as well as connecting and disconnecting tools. In order to reduce potential operational errors, as well as to decrease the operating time, the CRD100 was designed to automate these drilling tasks with fail-safe logic. The NASA Standard Reference Model for Telerobotics (NASREM) was chosen as the basis for the automation due to its modular, flexible, and easily expandable architecture. As well, the CRD100 drill provides an efficient Human-Machine Interface (HMI) between the operator(s) stationed on the surface vessel and the remote drill. The HMI allows both autonomous and manual control of the drill and displays a variety of real-time measurement data for monitoring the drilling process. The HMI has also been designed to allow pairs of operators to seamlessly hand-off tasks between each other and to log all pertinent data for later analysis.