Adaptive Neuro-Fuzzy Network Enhanced Automatic Visual Servoing Algorithm for ROV Manipulators

Abstract

This paper presents research and development for achieving advanced ROV manipulation systems with vision based servo control capable of being operated by pilots with auto assist in the dynamic subsea conditions. Underwater inspection and intervention operations are performed by work-class ROVs equipped with robotic manipulators. A standard offshore oil and gas setup includes a human pilot utilising telemanipulation technology to operate both vehicle and manipulators based on the work-site visual feedback provided by camera and sonar systems. For challenging applications in waves or currents where target devices are in motion a new approach is required. A position based visual servoing (PBVS) algorithm designed to follow a moving target with an underwater manipulator is proposed. The developed algorithm integrates Adaptive Neuro-Fuzzy Inference System (ANFIS) network framework for target motion prediction. The effectiveness of the developed software is verified through a series of experiments carried out with an off-the-shelf industrial hydraulic subsea manipulator in the laboratory conditions.