Hans Thomas

A deliberative architecture for AUV control

Autonomous Underwater Vehicles (AUVs) are an increasingly important tool for oceanographic research demonstrating their capabilities to sample the water column in depths far beyond what humans are capable of visiting, and doing so routinely and cost-effectively. However, control of these platforms to date has relied on fixed sequences for execution of pre-planned actions limiting their effectiveness for measuring dynamic and episodic ocean phenomenon. In this paper we present an agent architecture developed to overcome this limitation through on-board planning using Constraint- based Reasoning. Preliminary versions of the architecture have been integrated and tested in simulation and at sea.

Performance of an AUV navigation system at Arctic latitudes

In October of 2001 MBARI operated an AUV in the Arctic at latitudes exceeding 80/spl deg/. The navigation instruments consisted of a ring-laser gyro INS coupled with a DVL and GPS, a separate fiber-optic based gyrocompass, and a traditional flux-gate AHRS system. This paper describes the performance of these instruments at high latitudes.In October 2001, the Monterey Bay Aquarium Research Institute (MBARI) operated an autonomous underwater vehicle (AUV) in the Arctic at latitudes exceeding 80/spl deg/. The navigation instruments consisted of a ring-laser gyro inertial navigation system (INS) coupled with a DVL and GPS, a separate fiber-optic-based gyro-compass, and a traditional flux-gate AHRS system. The instruments were tested on deck, in open water, and under ice. This paper describes the performance of these instruments at high latitudes.

Eruptive and tectonic history of the Endeavour Segment, Juan de Fuca Ridge, based on AUV mapping data and lava flow ages

High-resolution bathymetric surveys from autonomous underwater vehicles ABE and D. Allan B. were merged to create a coregistered map of 71.7 km2 of the Endeavour Segment of the Juan de Fuca Ridge. Radiocarbon dating of foraminifera in cores from three dives of remotely operated vehicle Doc Ricketts provide minimum eruption ages for 40 lava flows that are combined with the bathymetric data to outline the eruptive and tectonic history. The ages range from Modern to 10,700 marine-calibrated years before present (yr BP). During a robust magmatic phase from >10,700 yr BP to ∼4300 yr BP, flows erupted from an axial high and many flowed >5 km down the flanks; some partly buried adjacent valleys. Axial magma chambers (AMCs) may have been wider than today to supply dike intrusions over a 2 km wide axial zone. Summit Seamount formed by ∼4770 yr BP and was subsequently dismembered during a period of extension with little volcanism starting ∼4300 yr BP. This tectonic phase with only rare volcanic eruptions lasted until ∼2300 yr BP and may have resulted in near-solidification of the AMCs. The axial graben formed by crustal extension during this period of low magmatic activity. Infrequent eruptions occurred on the flanks between 2620–1760 yr BP and within the axial graben since ∼1750 yr BP. This most recent phase of limited volcanic and intense hydrothermal activity that began ∼2300 yr BP defines a hydrothermal phase of ridge development that coincides with the present-day 1 km wide AMCs and overlying hydrothermal vent fields.

Volcanic morphology of West Mata Volcano, NE Lau Basin, based on high‐resolution bathymetry and depth changes

High-resolution (1.5 m) mapping from the autonomous underwater vehicle (AUV) D. Allan B. of West Mata Volcano in the northern Lau Basin is used to identify the processes that construct and modify the volcano. The surface consists largely of volcaniclastic debris that forms smooth slopes to the NW and SE, with smaller lava flows forming gently sloping plateaus concentrated along the ENE and WSW rift zones, and more elongate flows radiating from the summit. Two active volcanic vents, Prometheus and Hades, are located ∼50 and ∼150 m WSW of the 1159 m summit, respectively, and are slightly NW of the ridgeline so the most abundant clastic deposits are emplaced on the NW flank. This eruptive activity and the location of vents appears to have been persistent for more than a decade, based on comparison of ship-based bathymetric surveys in 1996 and 2008–2010, which show positive depth changes up to 96 m on the summit and north flank of the volcano. The widespread distribution of clastic deposits downslope from the rift zones, as well as from the current vents, suggests that pyroclastic activity occurs at least as deep as 2200 m. The similar morphology of additional nearby volcanoes suggests that they too have abundant pyroclastic deposits.

A peak-capture algorithm used on an autonomous underwater vehicle in the 2010 Gulf of Mexico oil spill response scientific survey

During the Gulf of Mexico Oil Spill Response Scientific Survey on the National Oceanic and Atmospheric Administration Ship Gordon Gunter Cruise GU-10-02 (27 May–4 June 2010), a Monterey Bay Aquarium Research Institute autonomous underwater vehicle (AUV) was deployed to make high-resolution surveys of the water column in targeted areas. There were 10 2-liter samplers on the AUV for acquiring water samples. An essential challenge was how to autonomously trigger the samplers when peak hydrocarbon signals were detected. In ship hydrocasts (measurements by lowered instruments) at a site to the southwest of the Deepwater Horizon wellhead, the hydrocarbon signal showed a sharp peak between 1,100- and 1,200-m depths, suggesting the existence of a horizontally oriented subsurface hydrocarbon plume. In response to this finding, we deployed the AUV at this site to make high-resolution surveys and acquire water samples. To autonomously trigger the samplers at peak hydrocarbon signals, we modified an algorithm that was previously developed for capturing peaks in a biological thin layer. The modified algorithm still uses the AUVs sawtooth (i.e., yo-yo) trajectory in the vertical dimension and takes advantage of the fact that in one yo-yo cycle, the vehicle crosses the horizontal plume (i.e., the strong-signal layer) twice. On the first crossing, the vehicle detects the peak and logs the corresponding depth (after correcting for the detection delay). On the second crossing, a sampling is triggered when the vehicle reaches the depth logged on the first crossing, based on the assumption that the depth of the horizontal oil layer does not vary much between two successive crossings that are no more than several hundred meters apart. In this paper, we present the algorithm and its performance in an AUV mission on 3 June 2010 in the Gulf of Mexico. In addition, we present an improvement to the algorithm and the corresponding results from postprocessing the AUV mission data.

Reliability growth of autonomous underwater vehicle-Dorado

Underwater environments are highly unstructured, uncertain, and dynamic. However, over the past three decades many autonomous underwater vehicles (AUVs) have been developed and successfully deployed for various oceanographic applications. There is a growing need for the AUVs to be reliable for collecting useful data and samples for their users. An AUV is a typical one-of-a-kind product for which it is difficult to derive a mathematical/statistical model for reliability prediction. The failure modes effects and criticality analysis (FMECA) can produce more effective results. However, the assessment of reliability growth of an AUV is very important for predicting its operational success in challenging underwater environments. In this paper, we present the identification of different types of failures that occurred during the past one and a half years of operation of MBARIs Dorado AUV, classification of those failures, and the reliability growth analysis for the vehicle. Reliability issues of various subsystems and operational procedure of the AUV have also been discussed. An extensive analysis of operational data and test results are presented in this paper.

Mapping payload development for MBARI's Dorado-class AUVs

The Monterey Bay Aquarium Research Institute (MBARI) is developing an autonomous seafloor mapping capability for deep ocean science applications. The MBARI Mapping AUV is a 0.53 m (21 in) diameter, 5.1 m (16.7 ft) long, Dorado-class vehicle designed to carry four mapping sonars. The primary sensor is a 200 kHz multibeam sonar producing swath bathymetry and sidescan. In addition, the vehicle carries 100 kHz and 410 kHz chirp sidescan sonars, and a 2-16 kHz sweep chirp subbottom profiler. Navigation and attitude data are obtained from an inertial navigation system (INS) incorporating a ring laser gyro and a 300 kHz Doppler velocity log (DVL). The vehicle also includes acoustic modem, ultra-short baseline navigation, and long-baseline navigation systems. A single cylindrical pressure housing contains all of the mapping sonar electronics, and the main vehicle control and acoustic communications electronics are housed in a separate glass ball. The Mapping AUV is powered by three 2 kWhr Li-polymer batteries, providing an expected mission duration of 12 hours at a typical speed of 1.5 m/s. The assembled package is rated to 6000 m depth, allowing MBARI to conduct high-resolution mapping of the deep-ocean seafloor. Initial at-sea testing commenced in May 2004 using the subbottom profiler and 100 kHz sidescan. The sonar package will also be mountable on ROV Ventana, allowing surveys at altitudes < 10 m at topographically challenging sites. The MBARI Seafloor Mapping team is now working towards integration of the multibeam sonar and towards achieving regular operations during 2005.

Preliminary Results for Model-Based Adaptive Control of an Autonomous Underwater Vehicle

We discuss a novel autonomous system which integrates onboard deliberation with execution and probabilistic state estimation for an adaptive Autonomous Underwater Vehicle for deep sea exploration. The work is motivated by the need to have AUVs be goal-directed, perceptive, adaptive and robust in the context of dynamic and uncertain conditions. The challenges leading to deployment required dealing with modeling uncertainty and integrating control loops at different levels of abstraction and response for a dynamic environment. The system is general-purpose and adaptable to other ocean going and terrestrial platforms.

Seafloor geomorphic manifestations of gas venting and shallow subbottom gas hydrate occurrences

High-resolution multibeam bathymetry data collected with an autonomous underwater vehicle (AUV) complemented by compressed high-intensity radar pulse (Chirp) profiles and remotely operated vehicle (ROV) observations and sediment sampling reveal a distinctive rough topography associated with seafloor gas venting and/or near-subsurface gas hydrate accumulations. The surveys provide 1 m bathymetric grids of deep-water gas venting sites along the best-known gas venting areas along the Pacific margin of North America, which is an unprecedented level of resolution. Patches of conspicuously rough seafloor that are tens of meters to hundreds of meters across and occur on larger seafloor topographic highs characterize seepage areas. Some patches are composed of multiple depressions that range from 1 to 100 m in diameter and are commonly up to 10 m deeper than the adjacent seafloor. Elevated mounds with relief of >10 m and fractured surfaces suggest that seafloor expansion also occurs. Ground truth observations show that these areas contain broken pavements of methane-derived authigenic carbonates with intervening topographic lows. Patterns seen in Chirp profiles, ROV observations, and core data suggest that the rough topography is produced by a combination of diagenetic alteration, focused erosion, and inflation of the seafloor. This characteristic texture allows previously unknown gas venting areas to be identified within these surveys. A conceptual model for the evolution of these features suggests that these morphologies develop slowly over protracted periods of slow seepage and shows the impact of gas venting and gas hydrate development on the seafloor morphology.

A High‐Resolution Survey Of A Deep Hydrocarbon Plume In The Gulf Of Mexico During The 2010 Macondo Blowout

Monitoring and M A Record-Breakin Geophysical Mon Copyright 2011 b 10.1029/2011GM Following destruction of the Deepwater Horizon drilling rig, while unmitigated blowout from the Macondo well was ongoing, NOAA scientific response cruise GU-10-02 (27 May to 4 June 2010) employed coordinated ship and autonomous underwater vehicle (AUV) operations to locate and study deep hydrocarbon plumes. The ship hydrocast survey localized maximum optical signals of a deep plume, centered at ~1150 m depth, approximately 13 km southwest of the blowout. Deployed at this location, the AUV conducted a high-resolution survey of plume structure, which indicated small-scale topographic influences on plume transport. Maximum plume intensity was observed along the western slope of Biloxi Dome. The orientation of gradients in plume intensity relative to isobaths indicated flow from the dome slope onto the dome top. In terms of the relative proportions of major hydrocarbon groups, all plume samples southwest of the blowout exhibited similar composition. The chemical composition of the plume southwest of the blowout was significantly different from the composition of a weaker deep plume observed southeast of the blowout. Variation in optical signal from a colored dissolved organic matter (CDOM) fluorometer (FCDOM) explained up to 97% (median 88%) of the variance in the concentrations of individual hydrocarbon compounds. AUV data also showed that FCDOM was highly correlated with three other optical measurements (r > 0.97) and oxygen measurements (r = 0.95). The results provide unique perspective on small-scale dynamics of a deep plume and illustrate the potential for studying subsurface plumes of dispersed oil using AUVs with off-the-shelf sensors.