An autonomous underwater vehicle

D. Maddalena, Tecnomare SpA and M. Berta, CEOM SpA

Deepwater operations require suitable means to carry out inspection and maintenance on deep-sea installations. Robotic technologies for Autonomous Underwater Vehicles (AUVs) are becoming mature enough for offshore-industry applications.

ROVs will maintain their strong position in deepwater construction and maintenance, but many inspection tasks can now be regarded as feasible with AUVs. Further, using ROVs is more expensive than a comparable AUV due to ROV-related requirements such as vessel size, umbilical winch and personnel. This advantage becomes more evident in deepwater and large-area inspections, where weight / size of handling equipment, along with deploying / retrieving time, make ROV inspections more expensive.

However, apart from military applications, AUVs have thus far been used for relatively simple jobs such as open-sea surveys. In addition, it is noteworthy that some of them are correctly classified as untethered underwater vehicles, since they are designed for piloting from the surface via an acoustic link.

Project RAIS. The Robotic Auto-nomous Inspection System is the outcome of a collaborative research program managed by CEOM and developed with technical contributions from Tecnomare, Snamprogetti and Sasp Offshore Engineering. Financial support is provided by the Italian R&D Directorate and Eni. The purpose of the current phase of the RAIS project is to develop and implement sophisticated techniques to perform complex tasks such as sealine inspections. Its final aim is development and testing of an AUV capable of performing the following reference tasks:

   Sealine visual inspections and free-span gauging will be performed in detail. This job has been satisfactorily achieved using an ROV. However, as mentioned above, this entails high costs due to the supply vessel and related staff, as well as slow operative speed and long survey time due to direct operator observation of the video image for detecting critical events. Instead, RAIS will be deployed by a small vessel-of-opportunity with reduced crew; further, experienced operators will not be necessary thanks to the advanced Man / Machine Interface and autonomous vehicle control.

   Survey of sea bottom and sub-bottom in high resolution is needed to plan and construct submarine pipelines. This requires comprehensive characterization of both bathymetry and seafloor morphology, as well as sedimentological features of the first bottom layer. An AUV, traveling tens of meters above the seafloor, provides an excellent platform for the relevant sonar equipment.

Basic robotic technologies. The essential robotic elements for enabling autonomous systems to perform tasks have been developed and tested in the lab. At-sea tests are foreseen this year. These technologies – and six, related main features – are as follows:

1. An algorithm, based on Kalman filtering, takes care of integrating information from different sensors such as Doppler sonar, attitude sensors (i.e., fiber optic gyros), absolute positioning provided by acoustic systems, landmark locations along the sealine, etc. A sensor, based on laser profiling, has been developed for measuring the relative position of RAIS with reference to the sealine.
2. Scene recognition is based on statistical comparison between actual image features and an a priori-learned reference. Pictures coming from a front-mounted TV are fed to a computer fitted with software to perform automatic identification and positioning of sealine profiles and landmarks, e.g., sacrificial anodes. Video processing has already been tested with pictures taken on a real sealine in varying conditions, such as partly buried, covered with marine fouling, etc.
3. Obstacle detection and sizing uses the signal coming from look-ahead sonar. It is processed to provide an "occupancy grid" of the space around the vehicle. This space is split into "cells" that can be marked as either "empty" or "obstacle," with different reliability levels. Echoes taken later from different vehicle positions are compared and integrated with a statistical approach to increase obstacle-detection reliability and avoid false alarms.
4. Supervisory-control software for online trajectory-planning software has been developed and validated to follow either the sealine or sea bottom, as accurately as possible, and to perform optimal (i.e., with minimum energy consumption) maneuvers for obstacle avoidance. Essentially, this algorithm evaluates, in real time, the best course; this is based on "attractive forces" due to the specified path (e.g., the sealine trajectory) and "repulsive forces" due to obstacles.
5. Due to their high energy / mass density, Ag-Zn cells have been selected for onboard energy storage.
6. The Man / Machine Interface will allow offline mission programing and online monitoring of mission execution.

It is worth noting that all of the above technologies are applicable to a broader range of purposes. For instance, in a subsea production scenario in ultra-deep water, a new approach can be implemented where an AUV navigates and docks to a Subsurface Production System (SPS). In this way, the vehicle can receive power from the SPS and communicate with the surface production facility – to which the SPS is connected – via the electrical umbilical. From this point on, the AUV can be tele-operated directly from the surface facility to perform tasks.