| dc.description.abstract | This thesis is about automatic motion control systems for remotely operated vehicles (ROV). The work has focused on topics within guidance and navigation. In addition, a motion control system has been developed, implemented, tested and used on two ROVs in sea trials.
The main motivation for the work has been the need to automate ROV tasks in order to make the ROV a more efficient tool for exploring the ocean space. Many parts of a motion control system for a ROV is similar to that of surface vessels and ships. However, there are many differences as well which require special solutions. This thesis proposes solutions to some of these challenges.
Underwater navigation is one of the main challenges of developing a motion control system as no global positioning system (GPS) is available below the surface. The work here contains contributions in both model-based and sensor-based state estimation methods. Model-based Kalman filters, both linearized and extended, have been developed and tested successfully on ROVs in sea trials. However, the need for an observer that works during manipulation work, and other cases with uncertain and rapid varying disturbances and dynamics, inspired the research on sensor-based state estimation. Specifically, a new type of attitude estimator known as an explicit complementary filter has been adopted and modified for ROV use. This attitude estimator is also used in an integration filter to estimate the translational positions. The main contribution here is a new method to include velocity measurements from a Doppler velocity log (DVL), or velocity estimates, to approximate the proper acceleration of the vehicle. This is used to improve the attitude estimation for accelerated vehicles, and hence the estimated positions when used in cascade with an integration filter.
Guidance of the ROV is needed to perform automatic tasks such as trajectory tracking and terrain following. The guidance modes require different levels of operator interaction. The thesis contains contributions on joystick in closed-loop control, reference models for A to B moves and path tracking, and altitude control and terrain following.
Joystick in closed-loop control compensates for dynamical and environmental forces in order for the pilot to easily guide the ROV. This work uses a filter-based reference model to generate desired trajectories based in the joystick commands. These trajectories are tracked by a closed-loop controller. The reference model is modified to reduce pilot induced oscillations. The human-in-the-loop is a special case where the interaction between the operator and the controller determines the overall performance of the ROV operation.
A simple and intuitive reference model for A to B moves and path tracking is a useful tool for surveys of the sea floor. This thesis proposes a constant jerk reference model. The idea is to generate feasible desired velocities and positions from integration of a constant jerk (time derivative of acceleration) that is on or off. Some benefits over filter-based reference models are that the maximum velocity is low and acceleration and deceleration stages are short. This gives a predictable behavior of the ROV for the pilot who is monitoring the operation.
In order to perform video surveys or photomosaicking of the sea floor, the ROV must be able to follow the terrain automatically. A new method for sea floor geometry approximation and altitude control based on DVL range measurements is proposed. The method has also been implemented in the developed motion control system and has been successfully used on two ROVs. The main contributions here are the sea floor geometry approximation method and a new guidance law for desired depth corresponding to the reference altitude. Thus, altitude control is obtained by using the depth controller. The main new contribution in the guidance law is a feed forward term that assures proper heave velocity.
As part of the work with the thesis, a motion control system, including software and hardware, has been developed in collaboration with other PhD candidates and MSc students, where the author has had the responsibility of putting every thing together under the supervision of postdoc Martin Ludvigsen and prof. Asgeir J. Sørensen. This motion control system has been implemented on two ROVs and it has been used in cruises for ocean sciences and pipeline inspection. | nb_NO |
