ADAPTIVE FUZZY CONTROL FOR UNDERWATER HYDRAULIC MANIPULATORS

Abstract

Underwater hydraulic manipulators are usually systems hard to be modeled and present strong non-linearities in its dynamics behavior. These types of manipulators are operated, nowadays, in a master-slave configuration with simple control algorithms performing tasks in hazardous and unstructured environments. In such conditions only low accuracy simple tasks can be performed. This paper presents the application of a special fuzzy controller in a hydraulic manipulator commonly used in offshore missions. This controller is able to cope with changes in the system parameters adapting itself to a desired system performance. A non-linear analysis of the manipulator's dynamics is developed. Results of simulations comparing the adaptive fuzzy control and a conventional control are also provided. Keywords. fuzzy control, adaptive control, hydraulic manipulators, underwater. Underwater manipulators mounted on remotely operated vehicles (ROVs) have an important role to play in deepwater activities for the oil and gas industry, in military and exploration missions. Nowadays, the use of these manipulators is based mainly in the concept of teleoperation using a master-slave configuration. The operator controls the manipulator (slave) moving a smaller master arm installed in a safe location distant to the worksite. A video cam mounted near the slave manipulator provides images of the work environment allowing the operator to realize the tasks. However, teleoperation present drawbacks that reduce its efficiency. The operator's work is made difficult by the poor quality of the video image and lack of sensibility of what's really happening in the remote location, permitting him to perform only relatively simple tasks. One step forward in teleoperation would be the accomplishment of tasks in a teleassisted way. This would allow the manipulator to perform tasks like positioning or trajectory following automatically, only supervised by the operator. However in the way to do that it's necessary to introduce a more sophisticated control than the one currently used since the manipulator operates in hazardous and unstructured environments and it's submitted to unknown and continuously changing conditions, that simple controllers are incapable of deal. Teleoperation avoids this problem permitting the operator to adjust the errors induced by the poor controller observing the slave arm with the cam and adjusting its position moving the master arm. Another difficult arises from the actuation system used by underwater manipulators; most of them are hydraulic actuated because of the robustness and good relation force/weight. This also means a further trouble to the control because its dynamics has non-linearities and factors that difficult an accurate positioning of the actuator, like compressibility of the fluid and leakage in the piston's seals. Recently there is an effort to adapt industrial electrical manipulators to operate underwater, but we are still far away to see them working. The purpose of this paper is to simulate the application, in an underwater hydraulic manipulator, of a control capable to deal with the non-linearities of a hydraulic actuated system and also able to coping with changes in the system's parameters, always maintaining a suitable performance. The control here utilized is a fuzzy control referenced by Layne (1996) as Fuzzy Model Reference Learning Control (FMRLC). In the last years fuzzy controllers has demonstrated to be an effective alternative to conventional controllers, especially when one is interested in controlling systems with non-linearities and/or hard to be modeled. In fact there are many successful cases of fuzzy applications, like air conditioning systems, anti-lock brakes, chemical mixer, elevator control, camera autofocus, and even on robotics systems. Sepehri (1999) already showed a fuzzy control applied in a hydraulic actuated industrial robot. This papers goes further, showing as a simple but powerful algorithm