Guide and Control of an unmanned underwater Vehicle by Visual-feedback techniques

Manual guidance and control of underwater vehicles, like ROVs, is in general a demanding task. A great help comes, in this situation, from the development of automatic guidance systems that, taking care of low and mid level control task, may reduce the burden imposed to human operators. A system of this kind has been constructed for guidance and control of a small work-class ROV (DOE Phantom S2) by inserting a PC in the operator/ROV loop. The standard configuration of the control loop turns out, in this way, to be completely redesigned as a PC-based (control) configuration. The paper describes the basic features of the realized system, that, by allowing both on-line processing of sensory data and implementation of automatic control procedures, enhances the performances of the ROV as a robotic tools. As a specific application of the PC-based control configuration, the implementation of a visual feedback control strategy in a typical inspection mission (underwater pipeline’s inspection) is considered. Images from a CCD camera are processed at high logical level in order to extract information about the relative ROV/pipe position. This information is in turn used for generating and feeding back suitable commands to the ROV’s thrusters, in order to regulate the vehicle position. The system has undergone both field tests and emulation tests, showing a satisfactory behavior that is illustrated and discussed.