Search

This thesis describes a course of work aimed at developing practical methods of implementing compliant force control to enable the automation of sub-sea inspection of existing offshore structures. The overall system is based on a six degree-of-freedom (dof) hydraulic manipulator mounted on a two dof platform which is mounted on a remotely operated vehicle (ROV). For inspection tasks, the ROV attaches itself to the structure using three vacuum feet to maintain a reasonably stiff working platform. The manipulator may then be deployed to surface track the weld of two or more intersecting structural members whilst holding the appropriate inspection probe. For this project an Alternating Current Field Measurement (ACFM) probe is used. This does not need electrical surface contact but instead can operate with a constant elevation above a surface provided by the probe's own casing. This thesis also considers the likely motions that will be experienced by the real system and examines the use of vision systems to compensate for unwanted movement. It further examines the design of an appropriate compliance to sit between the manipulator and the probe and describes the construction of this compliance device. A novel method is offered that uses an artificial neural network (ANN) to simplify the problem of mapping a robot tool position in space when using compliant force control. For this case, the compliant device is a network of springs between plates situated between the manipulator end flange and the tool. A force sensor is mounted between the tool and the compliant device. To overcome the difficulties of accurately modelling the non-linear and highly coupled characteristics of the compliant device, a back propagating ANN was trained to relate the forces to positions for real data sets. The finished ANN was used as part of a computer model of the overall manipulator system. The ANN was then used as part of a force control loop to enable surface tracking.