From Test Tank to Deep Water01/01/1970 |
| Sub-sea inspection using autonomous underwater vehicle (SPINAV) |
| SPINAV is an ongoing Joint Industry Project that is moving cutting-edge autonomous inspection technology from proof-of-concept trials in the test tank (completed November 2004) to an offshore trial using an unmodified commercial work-class ROV or AUV of opportunity. This article reviews the background to the technology and discusses details of the development process and wet trials. |
| Dr Kelvin Hamilton and Dr Jonathan Evans, SeeByte Ltd, Scotland |
The initial part of this ongoing project, SPINAV-1, the proof-of-concept trial had several sponsors: BP, Subsea7, Conoco-Philips and the UK Government Department of Trade and Industry. SPINAV-2, the offshore work-up project, is now underway; started in July 2005 and running for twelve months, it is sponsored by BP and Scottish Enterprise. Both these JIPs were organised through the ITF (Industry Technology Facilitator). Current practice for most forms of offshore inspection is to use an ROV equipped with video camera and other sensors, such as AC Impedance sensor, with forward-looking sonar for localisation in poor visibility. Occasionally, towed side-scan can be used (e.g. pipelines) for low-fidelity information.
The CCMM module sits above all other embedded vehicle modules and is the interface between the pilot’s SeeTrack graphical mission planning tool and the vehicle. As most missions mainly involve moving from one place to another, the CCMM deals mainly with the Autopilot module. The CCMM module presently sends commands at 1-second intervals to the lower level ‘real-time’ modules such as the Autopilot. If an anomaly is detected the CCMM module will adapt its mission to perform a close video inspection of the area of interest. Once this has been performed the normal mission is resumed. In SPINAV-2 the CCMM module will receive an incremental upgrade to allow complex 3D manoeuvres and mission adaptation around generic structures, not just risers. Object tracking such as riser or wall tracking is performed by image processing performed in real time upon various types of sensor data. For SPINAV-1 the Didson acoustic camera was used, managed by a driver module that takes care of setting the correct ranges and operating modes. The raw Didson image is then employed by the Tracker module, which uses image-processing techniques to recognise any riser in the image and determine range and bearing to the riser centre with respect to the vehicle. For SPINAV-2 a combination of other tracking-suitable sensors will be explored. An autopilot monitors vehicle position and responds to movement requests by calculating the necessary thruster settings to achieve this position. It adjusts thrusters at five times a second, which is generally enough to control the relatively slow ROVs or AUVs in use today. This provides a very smooth, precise and efficient level of vehicle control and is markedly different to that produced by a human ROV pilot. In SPINAV-1 the autopilot had to be manually tuned to suit the vehicle under control, but SPINAV-2 will see the addition of an ‘autotune’ facility, allowing the autopilot to take control of different vehicles without a long tuning period. Visualisation of the mass of inspection data is performed using SeeByte’s SeeTrack system. In SPINAV-1 vehicle track, video data and fluorosene concentration were combined into a single, easy-to-use 3D world. Using this it was possible to immediately see any dye in the water, click on the point in question and view a video picture of that area. This simplified the entire analysis task for the operators. The developmental platform chosen for the sheltered-water trials of SPINAV is Heriot-Watt University’s Rauver AUV, a <250kg AUV capable of hovering, inspection and light intervention, as well as more conventional survey missions. Open construction and ease of interfacing enable it to be used as a convenient test-bed for various payloads. It can function either as AUV or as a battery-operated ROV. Demonstrating Skills During SPINAV-1 a live demonstration was given to various oil, gas and military representatives. The objective was to inspect an 8m-tall, 13cm-diameter riser which had a transparent bottle of fluorosene dye representing a plume attached at an arbitrary point. The SPINAV system running on the Rauver AUV dived and began circling the riser until it saw the bottle of dye with its fluorosene sensor. At this point, the CCMM commanded a close inspection of the area of interest before resuming its normal inspection mission. At the end of the inspection, which took about 15 minutes in total, Rauver surfaced and the data was downloaded into SeeTrack via Rauver’s surface WiFi connection. Within 20 minutes of Rauver surfacing, the operator was able to show observers the entire mission in 3D, together with fluorosene-concentration maps and close-inspection video footage of the dye bottle. A screenshot of the SeeTrack view of the demonstration is shown in Figure 4. In SPINAV-2 this will be extended to allow full 3D texture-mapped mosaicking of video and Didson data, together with indications of any coverage gaps. An early version of this is shown in Figure 5, showing a 3D view of a vehicle, its track and some 3D structures waiting for video or Didson data to be mosaicked onto them. |
| Biography of the author Dr Kelvin Hamilton is the Robotics Development Manager for SeeByte Ltd. He worked in the manufacturing industry for many years before gaining his Ph.D. in Autonomous Robotics at Heriot-Watt University, Edinburgh, where he designed the original RAUVER AUV as a testbed platform. He is currently responsible for the SPINAV autonomous inspection programme. |
|
| Archive |
| Archive > September 2005, Volume 9, Number 7 > From Test Tank to Deep Water |
|
