Nereus hybrid underwater robotic vehicle - IngentaConnect [PDF]

The Nereus hybrid underwater robotic vehicle AD Bowen, DR Yoerger, C Taylor, R McCabe, J Howland, D Gomez-Ibanez, JC Kinsey, M Heintz, G McDonald and D Peters Department of Applied Ocean Physics and Engineering, Woods Hole Oceanographic Institution, Woods Hole, MA, USA

Technical Paper

doi:10.3723/ut.28.079 International Journal of the Society for Underwater Technology, Vol 28, No 3, pp 79–89, 2009

C Young, J Buescher and B Fletcher US Navy Space and Naval Warfare Systems Center, San Diego, CA, USA LL Whitcomb, SC Martin, SE Webster and MV Jakuba Department of Applied Ocean Physics and Engineering, Woods Hole Oceanographic Institution, Woods Hole, MA, USA Department of Mechanical Engineering, The Johns Hopkins University, Baltimore, MD, USA

Abstract The Nereus vehicle will enable scientists to explore remote regions of the oceans, such as under the polar ice caps and deep trenches, up to depths of 10 972m (36 000ft). Technology limitations have prevented routine, cost-effective access to these remote regions, and the final 4500m of the ocean remain largely unexplored. New solutions to deep diving are described. The Nereus hybrid remotely operated vehicle (HROV) is designed for exploration and research needs as a single system. It can operate as an autonomous vehicle for seafloor surveys, or in a tethered/ROV mode to sample rocks or deep-sea animals The HROV Nereus transforms between its two modes of operation to accomplish all these tasks during a single cruise deployment. Sea trials of Nereus took place off the Hawaiian Islands at 2500m in November 2007. An overview of the vehicle and results from its initial trials are reported here. Keywords: autonomous intervention, hybrid ROV/AUV, Marianas Trench, ceramic, LED lighting, Nereus, hybrid remotely operated vehicle (HROV)

1. Introduction Existing deep submergence vehicle systems have excellent capabilities and provide critical, routine access to the seafloor to a maximum depth range of 4000–6500m – e.g. the 4500m Alvin humanoccupied submersible (Broad, 1997; Kaharl, 1990), the 4500m Autonomous Benthic Explorer (ABE) autonomous underwater vehicle (AUV) (Yoerger et al., 1998, 2007) and the 4000m Tiburon remotely operated vehicle (ROV) (Newman and Stakes, 1994). Only one presently operational US vehicle is capable of diving to 6500m and conducting

high resolution mapping and sampling: the 6500m Jason II ROV (Whitcomb et al., 2003). Progress in deep-sea research at ocean floor sites between 6500 and 11 000m has been hindered by a lack of suitable cost-effective vehicles that can operate at these depths. Given the need for full access to the global abyss, and national and international imperatives regarding ocean exploration, several studies have identified the development of an 11 000m deep submergence vehicle as a US national priority (National Science Foundation [NSF], 1996; University-National Oceanographic Laboratory System [UNOLS], 2000; National Oceanic and Atmospheric Administration [NOAA], 2000; Shepard et al., 2002). To date, only two vehicles have ever reached the deepest place on earth, which is Challenger Deep of the Marianas Trench at 11◦ 220 N, 142◦ 250 E in the western Pacific Ocean near the island of Guam (Fryer et al., 2003). On 23 January 1960, the human-piloted bathyscaphe Trieste, developed by Auguste Piccard, made one successful dive to the Challenger Deep (Piccard and Dietz, 1961). In 1995 the ROV Kaiko, built and operated by the Japan Agency for Marine Earth Science and Technology (JAMSTEC), made the first of several successful dives to the Challenger Deep (Takagawa, 1995). Neither Trieste nor Kaiko is presently operational. Moreover, the design approaches employed in these two very different vehicles necessarily resulted in high operational costs – too costly to be supported by US oceanographic science budgets. More recent employment of simple bottom-landers (Jamieson et al., 2008) has provided cost-effective access to these hadal depths and has underscored the dearth of information about various processes below 6000m.

79

Bowen et al. The Nereus hybrid underwater robotic vehicle

FIBER PAYOUT FROM ARMORED CABLE DEPRESSOR

ARMORED CABLE DEPRESSOR

ARMORED CABLE DEPRESSOR

VEHICLE

FIBER PAYOUT FROM VEHICLE AS NEEDED

FIBER PAYOUT FROM VEHICLE AS NEEDED

1

2

VEHICLE TRAVELS TOWARD TARGET

RETRIEVED FIBER WOUND ONTO RECOVERY WINCH

VEHICLE WORKS ON BOTTOM

3

VEHICLE AND ARMORED CABLE DEPRESSOR ABROAD

VEHICLE RETURNS TO SHIP

FIBER END DROPPED FROM VEHICLE

4

FIBER END DROPPED FROM VEHICLE

5

Fig 1: Nereus ROV mode concept of operations: in ROV mode, Nereus is remotely controlled by a

lightweight, expendable, fibre-optic tether that connects the vehicle to a surface support vessel Light fibre-optic tethers offer an alternative to conventional large-diameter tethers. McFarlane (1990) imagined a deep-diving vehicle Cybernaut utilising a small-diameter tether to trickle-charge onboard batteries. To date, however, light fibre tethers have principally been employed in military applications; relatively few light fibre tether systems have been employed for oceanographic research. Aoki et al. (1992) and Murashima et al. (1999) reported the development of the self-powered UROV7K employing a fibre-optic tether. This vehicle was designed to operate exclusively as a tethered ROV and did not have onboard computational resources necessary to operate autonomously. Our goal is to create a practical 11 000m system using an appropriately designed self-powered vehicle that can (a ) operate as an untethered autonomous vehicle in AUV mode and (b ) operate under remote control connected to the surface vessel by a lightweight fibre-optic tether of up to approximately 40km in length in ROV mode (Bowen et al., 2004). Fig 1 depicts the tethered ROV mode concept of operations. Fig 2a shows Nereus in its AUV mode and Fig 2b shows it in ROV mode. This paper reports an overview of the new Nereus hybrid underwater vehicle and summarises

80

its performance during its first sea trials. For broad area survey, the vehicle can operate untethered as an AUV capable of exploring and mapping the seafloor with sonars and cameras. For close-up imaging and sampling, Nereus can be converted at sea to become a tethered ROV. The ROV configuration incorporates a novel lightweight fibre-optic tether to the surface for high bandwidth real-time video and data telemetry to the surface, enabling high-quality remote-controlled teleoperation by a human pilot. Nereus ’ first sea trials were conducted in November 2007 from the RV Kilo Moana in the Pacific Ocean near Oahu, Hawaii. The sea trials demonstrated vehicle operations in both AUV and ROV modes to a depth of 2270m. Table 1 summarises the Nereus sea trial dive statistics. Future sea trials in May 2009 are planned to demonstrate the vehicle’s capability for operations at depths to 11 000m and with extreme horizontal mobility. The Nereus vehicle project lead institution is the Woods Hole Oceanographic Institution (WHOI) with collaboration of the Johns Hopkins University and the US Navy Space and Naval Warfare Systems Center San Diego.

Vol 28, No 3, 2009

a

b

Fig 2: The Nereus HROV, shown during its first engineering sea trials in the Pacific in November 2007, is

designed to operate in two modes to depths of 11 000m: (a) Nereus configured for AUV survey operations and (b) Nereus configured with a light fibre-optic tether, a robot arm, sampling gear and additional cameras for teleoperation of close-up imaging, sampling and manipulation missions Table 1: Dive statistics – November 2007 Nereus sea trails Date

Dive

Vehicle mode

Time submerged

Distance covered (m)

Depth (m)

Time on bottom

Fibre payout

39406 39407 39409 39409 39410 39410 39411 39412

Freejoy NER000 NER001 NER002 NER003 NER004 NER005 NER006

AUV AUV ROV ROV ROV ROV ROV AUV

5hr 56min 2hr 12min 57min 5hr 56min 1hr 20min 3hr 47min 6hr 50min 11hr 33min

3940 2319 0 1751 0 2236 2270 10843

2 4 18 398 100 569 2257 22

0min 0min 0min 4hr 55min 0min 2hr 43min 3hr 21min 12min

No tether No tether 28m 878m 1177m 744m 2380m No tether

2. Hybrid vehicle design overview The Nereus core vehicle employs twin free-flooded hulls. A tandem hull design was chosen to accommodate both operating modes of the vehicle. All onboard electronics, batteries and internal sensors are housed at 1 atmosphere in novel lightweight ceramic/titanium pressure housings developed specifically for this project (Stachiw et al., 2006). Additional buoyancy is provided by lightweight hollow ceramic buoyancy spheres (Stachiw and Peters, 2005; Weston et al., 2005). Nereus ’ power is provided by an 18kWh lithium-ion battery pack, developed for this project, which is contained in two ceramic pressure housings. Two 0.355m outside-diameter (OD) ceramic pressure housings contain power switching and distribution systems, DC–DC power isolation, a Linux control computer, a Linux imaging computer, DC-brushless motor amplifiers, multiple gigabit Ethernet transceivers, strap-down navigation sensors, an external sensor and actuator interfaces. Cameras, emergency beacons, RF modem (for surface operations) and other electronics are housed separately in dedicated 0.191m OD ceramic and titanium pressure housings. Lighting is provided by lightweight, ambient-pressure lightemitting diode (LED) arrays custom developed for the Nereus project (Howland et al., 2006). Pressure-balanced, oil-filled junction boxes and

hoses provide vehicle electrical and optical interconnect. The vehicle’s large metacentric height provides passive stability in roll and pitch. Twin aft vertical stabilisers provide passive hydrodynamic stability in heading.

2.1. Nereus AUV mode configuration summary In AUV mode, Nereus is neutrally buoyant with a displacement of 2500kg with 1510 ceramic buoyancy spheres and a reserve payload buoyancy of 22kg. This mode employs two independently articulated, actively controlled foils (wings) located between the hulls at the aft and middle sections, respectively. AUV mode propulsion is provided by two 1kW thrusters fixed on the aft tails and one 1kW thruster on the articulated mid-foil. AUV mode has no lateral thruster actuation. The vehicle is hydrodynamically stable in pitch and heading when in forward flight. A downward-looking survey camera and several LED arrays are mounted on the port hull. Additionally, a scanning sonar can collect high resolution bathymetry of the seafloor. 2.2. Nereus ROV mode configuration summary In ROV mode, Nereus is neutrally buoyant with a displacement of 2700kg with 1670 ceramic buoyancy spheres and a reserve payload buoyancy of 45kg. This mode adds a work package containing a 6◦ of freedom (6DOF) electrohydraulic robot arm, 81

Bowen et al. The Nereus hybrid underwater robotic vehicle

Fig 3: Main vehicle flotation structure for one hull

(one of two) sampling tools, sample containers, an additional high resolution digital camera, two utility cameras and several LED arrays. ROV mode propulsion is provided by two 1kW thrusters fixed on the aft tails, as used in the AUV mode, with the addition of one lateral 1kW thruster and one vertical 1kW thruster.

2.3. Ceramic buoyancy spheres and housings Ceramic buoyancy spheres and custom ceramic– titanium pressure housings were essential towards minimising the vehicle overall size and mass. 2.3.1. Ceramic buoyancy spheres Ceramic flotation was selected because of its low weight-to-displacement characteristics. The ones chosen were 99.9% alumina ceramic seamless spheres of 91mm OD, manufactured by Deep Sea Power and Light (DSPL) in San Diego, California (Weston et al., 2005). Each sphere weighs 140g and displaces 404g in seawater, for a nominal 0.35 weight-to-displacement ratio (compared with 0.59 for a syntactic foam for the same working depth). The spheres supplied by DSPL were individually tested to 207MPa external pressure. They are individually jacketed with 5mm-thick elastomeric boots that provide robust protection against impact loading, while providing an additional 19g of net buoyancy. Each booted sphere produces 283g of buoyancy at the surface. The spheres are less compressible than water, so at full operating depth of 11 000m they generate 306g of buoyancy each. The main vehicle flotation consists of 1472 spheres that are arranged in the upper portions of the vehicle hulls and generate 417kg of net buoyancy. The spheres are housed in longitudinal tubes within buoyant polypropylene modules. The polypropylene modules are constructed as arrays of 25mm plate stock held together with aluminium tie rods. The plates are oriented vertically and have patterns of circular cut-outs that form tubular holes when the modules are assembled. The polypropylene structure supporting the main vehicle flotation has a total weight of 690kg and displacement of 778kg, for a net buoyancy of 88kg. Total buoyancy

82

Fig 4: A main ceramic pressure housing being

readied for fitting into Nereus generated by spheres and polypropylene structural modules (Fig 3) is 505kg at the surface. Due to the fact that water is more compressible than both the spheres and the plastic structure, the total buoyancy at 11 000m depth is 573kg. 2.3.2. Ceramic pressure housings Ceramic was selected for pressure housing material because its high compressive strength-to-weight ratio allows for near neutrally buoyant housings capable of going to extreme depths. Titanium housings of this size and quantity required for Nereus would weigh hundreds of kilograms in water and would require expensive, voluminous, additional flotation to offset their weight, increasing drag and requiring higher power propulsion, more batteries and other factors. The housings are composed of 99.6% alumina ceramic and were custom manufactured to specification for Nereus by CoorsTek (Fig 4). The housings consist of a ceramic section by CoorsTek and titanium joint rings, which were designed and manufactured at WHOI, bonded to the ceramic with a high strength epoxy. Used as housing closures are hemispherical titanium end-caps, provided with all necessary penetrations for electrical and optical conductors, as well as purge ports. Finite element analysis was performed to ensure the matching of deflections and stresses of the ceramic and titanium components under pressure, as well as even axial loading on the ceramic components. Strain gauging and acoustic emissions measurement were performed during pressure testing of all housings to a proof pressure of 124MPa (Stachiw et al., 2006).

2.4. Fibre-optic tether A key part of the HROV system is the lightweight fibre-optic cable used when operating in the

Vol 28, No 3, 2009

Table 2: Candidate fibre tether mechanical

Table 3: Nereus electrohydraulic manipulator

specifications

performance specifications

Fibre parameter

FOMC

Buffered fibre

Diameter Specific gravity (fresh water) Weight of 11km in water Working strength Breaking strength Survivability on seafloor

0.8mm 1.74 4.23kg 133N 400N Good

0.25mm 1.36 0.173kg 8N 108N Poor

ROV mode. The fibre-optic tether transmits high bandwidth data only, not power. The preliminary design analysis of this cable for deep-ocean deployments is described in Young et al. (2006). Two cable designs were selected as candidates for the HROV system: fibre-optic microcable (FOMC) and the Sanmina-SCI buffered fibre. Characteristics of these cables are described in Table 2. After extensive simulation using the WHOI Cable program (Gobat and Grosenbaugh, 2000, 2001), coupled with field trials, the Sanmina/SCI tether was selected as the primary choice for the HROV system. The basic concept of the deployment system involves the use of a snag resistant depressor and vehicle package to house the tether system, as shown in Fig 2. The depressor was designed to get the upper tether deployment point below both the surface currents and the most energetic and biologically active part of the water column. The vehicle package contains an optical fibre dispenser, brake, fibre counter and cutter, and it is designed to minimise drag and the chance of snagging the fibre. The depressor and vehicle package are mated together during launch, protecting the fibre during the transition through the air-water interface. Once the system has reached a designated depth, the vehicle package separates from the depressor. As the vehicle package descends, optical fibre pays out both from it and the depressor. The cable deployment system was integrated with the actual Nereus HROV vehicle in autumn 2007 and reported on in further detail in Fletcher et al. (2008). Operational procedures for launch and recovery were developed and tested. Four dives were made in the ROV configuration using the fibre, culminating in a 4.5-hour dive to 2267m. During all ROV dives, the fibre remained intact until purposely cut at the end of planned operations.

2.5. Nereus manipulator and sampling system The Nereus sampling system consists of a 2.43 × 1.21m platform with a custom designed Kraft TeleRobotics Inc manipulator and a WHOI designed hydraulic power unit. The manipulator

Manipulator parameter

Value

Horizontal reach Lift capacity at full extension Controllable grip closure force Weight in air (sea water) Electrical power input Operating depth Jaw opening (4-finger)

1.5m +14kg 0–450N 48kg (32kg) 100W to 750W (max) 11 000m 22cm

is a seven-function, 6DOF, closed-loop, position controlled system, in a master slave configuration. The kinematics were developed to maximise the Nereus workspace, as outlined in Table 3. The Nereus hydraulic power unit (HPU) that provides hydraulic pressure to the manipulator is a WHOI designed modular hydraulic system that delivers 7MPa at flow rates of 7.6L per minute. The HPU uses the same DC-brushless motor and controllers as the Nereus propulsion system. It has a small, lightweight gear pump and a proportional relief valve and operates in torque mode, thereby using minimal power at all times. The maximum power used at any time with the HPU and manipulator is 750 watts. The hydraulic fluid is a biodegradable mineral oil-based, International Organization for Standardization (ISO) 7, marine hydraulic fluid, developed specially for Nereus ’ extreme ambient pressures. The sampling platform provides storage facilities for science equipment and samples. It was developed in conjunction with the manipulator to maximise the kinematic capabilities of the manipulator system. The manipulator is placed on the platform such that it has a full 270◦ of reach on the starboard corner of Nereus within the camera views. This provides a useful reach area of approximately 5.5m2 of seafloor.

2.6. Cameras and lighting The power-limited nature of Nereus required careful consideration of the requirements for imaging and resulted in a non-traditional approach of tightly integrating the imaging and lighting system. Early in the system design process, it was concluded that conventional full-frame rate imaging was not necessary for many of Nereus ’ tasks, so this feature could be limited to reduce imaging and lighting power requirements. Minimisation of frame rates is particularly valuable in ROV mode, when high levels of light output are important to producing quality colour imagery. Such a system requires non-traditional lighting sources capable of tailoring their output to the needs of the imaging cameras. Newly available, high brightness LEDs offered a means to strobe at high rates, as well as other 83

Bowen et al. The Nereus hybrid underwater robotic vehicle

benefits such as pressure tolerance and precise control over lighting patterns. For ROV mode imaging, a full-motion-capable colour imager based upon a machine vision camera used in previous deep-sea work has been employed (Howland et al., 2006). The camera is capable of capturing a high resolution colour image whenever it is triggered, at rates up to 29.97 frames per second. The camera trigger causes LED strobe discharge with a selectable pulse length, allowing effective intensity to be controlled. The resulting system thus uses power to produce light of the required intensity and duration only when required. A topside computer collects and buffers the imagery so that the viewer only sees illuminated frames of imagery. Triggering at these varying frame rates and durations required the development of a custom LED driver controller board. The controller is instructed via software the expected frame rate, which is driven by the needs of the pilot and depends entirely upon the need for visual update about vehicle activities. In another attempt to save power, only areas viewed by the cameras are illuminated. A 16-LED assembly, or puck, was designed. The lighting array is composed of multiple, individually controlled pucks, each capable of illuminating a 30◦ frustum. By selecting the individual pucks to be triggered, the light can be directed to only areas requiring illumination. The LED controller is directed via software to trigger those individual pucks. The pucks are the result of a custom design effort and contain 16 LEDs in a reflector assembly carefully designed to create an even lighting pattern. They are fully pressure compensated and have been tested to 11 000m. Pressure compensation of the assemblies results in dramatic weight savings over traditional pressureresistant lighting enclosures. The Nereus design includes 18 strobing LED pucks arranged in an array on the nose of one of the hulls. Two other pucks, used in a continuous lighting mode (not strobed), are also included in the design as backups, in the event of controller failure. In addition to the custom motion camera, two conventional subsea video cameras in 11 000m housings support viewing of the work package and other utility tasks. Fig 5 shows a preliminary mounting of several of the pucks during the Nereus sea trials. All of the imagery, including the conventional video camera data, is brought to the surface for viewing and recording over gigabit Ethernet channels. Nereus AUV imaging is more conventional, utilising an array of eight of the same design pucks and the same LED controller, supporting cyan LEDs in support of a greyscale camera.

84

Fig 5: Nereus showing the float-pack separating

from the depressor pack 2.7. Power and propulsion Nereus is powered by an 18kWh rechargeable lithium-ion battery system providing a 50V bus. Building upon lithium-ion experience with the ABE and REMUS AUVs, WHOI designed a modular building block of 12 cylindrical cells. The process included a comprehensive safety analysis, followed by independent laboratory testing for conformance to United Nations shipping regulations. The 12-cell packs are assembled into two identical chassis on Nereus and combined electrically onto a common bus providing up to 3kW continuous for all systems. Power is distributed to low-power devices from both sides of the vehicle in order to minimise cross vehicle wiring. Both port and starboard chassis contain low-power switches that feed individual power conversion units. These units provide electrical isolation, allowing for individual ground fault detection and minimisation of ground loops, and are roughly sized according to the loads for maximum efficiency. High-power devices are few in number and are all switched from a single chassis. These include propulsion, control surface and lighting devices, each consuming up to 1kW. High-power switches are built around solid-state relays that are controlled by a custom interface circuit board, which includes soft start as well as voltage, current and temperature sensors. The propulsion system is the largest load on the power bus and therefore has the greatest impact on mission duration. Hydrodynamic modelling indicated the drag of the two-hull AUV geometry would be 230N at the target speed of 1.5m/s. The corresponding electrical energy fell within the overall capacity of the power system. AUV mission profile simulations, which estimated the energy used by the rest of Nereus systems, as well as propulsion, yielded reasonable AUV mission lengths. As a result of design choices favouring AUV propulsion, ROV manoeuvring is not fully

Vol 28, No 3, 2009

optimised. Vertical thrust is limited to 350N, so buoyancy must be monitored carefully and yaw rate is limited by the long swing of the aft thrusters. A lateral thruster is added with the work package, improving worksite manoeuvrability. The November 2007 sea trials demonstrated that Nereus in ROV mode is a power-efficient and capable sampling platform. During a dive on 25 November 2007, the vehicle was submerged for 6hrs and 50min, with 3hrs, 21min on bottom at 2270m depth. Successful manipulation exercises were performed on glass bottles, a can, a large piece of metal debris and several sediment push cores, all while covering 2200m in horizontal distance. Including pre-dive and recovery, 100.9Ah battery charge was used out of a total 369Ah available – approximately 27% of capacity.

2.8. Navigation Nereus ’ extreme operating depth presents challenges for navigation sensing and estimation. In 11 000m operations, the long acoustic paths of conventional long baseline (LBL) acoustic navigation gives rise to problems of signal attenuation, decreased accuracy and limited update rates (Hunt et al., 1974). These challenges motivate the need to combine LBL navigation with Doppler navigation (Whitcomb et al., 1999; Kinsey and Whitcomb, 2004) and inertial navigation (Larsen, 2000; Alameda, 2002; McEwen et al., 2005) in order to obtain the precise, geodetically referenced navigation necessary for closed-loop control and benthic science. The Nereus navigation sensor suite includes a Paroscientific pressure depth sensor, a Teledyne 300kHz Doppler sonar, an Ixsea PHINS inertial measurement unit (IMU), a LBL transceiver, an WHOI Micro-Modem (Freitag et al., 2005) and a Microstrain gyro-stabilised attitude and magnetic heading sensor. The Doppler sonar provides threeaxis bottom-lock vehicle velocity with respect to the seafloor at over 200m altitude, three-axis water-lock velocity of the vehicle with respect to the water and three-axis water-column velocity profiles. The IMU contains a three-axis north-seeking fibre-optic gyrocompass providing attitude and heading at 0.01◦ accuracy. Navigation sensor data is received by Nereus ’ control computer. NavEst, the navigation software developed by WHOI and Johns Hopkins University for use on deep submergence vehicles, computes the vehicle state estimates. Presently employed on the Sentry AUV and Nereus, NavEst is a multithreaded Linux program that supports multiple simultaneous navigation algorithms. Available navigation algorithms include the Doppler navigation algorithm employed by DVLNAV

Fig 6: Nereus dive NER005 – screenshot of the

DVLNAV program during dive NER005 (Kinsey and Whitcomb, 2004) and the LBL algorithm extensively used by the ABE AUV (Yoerger et al., 2007). Single-beacon one-waytravel time algorithms have also been implemented (Eustice et al., 2007). During the November 2007 engineering trials, Nereus real-time navigation employed a pressure depth sensor, PHINS IMU and Doppler sonar. Fixed seafloor transponders were not deployed, precluding the use of LBL navigation. The sensors and NavEst software provided vehicle-state estimates that enabled closed-loop trajectory tracking control in both ROV and AUV modes. In ROV mode, navigation data is broadcast via Ethernet to computers topside and visualised using the DVLNAV software package. In AUV mode, DVLNAV displays basic vehicle navigation and engineeringstate information provided by the acoustic modem. Fig 6 shows a screenshot of the DVLNAV program during dive NER005 at 2257m depth during closedloop trackline navigation and control testing.

2.9. Controls and dynamics in AUV mode When configured as an AUV, Nereus is designed to be efficient in forward flight at speeds up to 2m/s. Unlike most AUVs, Nereus can stop, hover and reverse direction if necessary to negotiate steep terrain. The ABE vehicle has demonstrated the value of this latter capability for seafloor scientific surveys (Yoerger et al., 1998). Unlike AUVs with static thruster configurations (ABE, SeaBED ) or more typical fin-actuated AUVs, however, Nereus ’ depth control strategy changes as a function of speed. This increased complexity comes with the substantial benefits of low drag during cruising and alignment of the thrusters with the predominant flow direction, except at very low speeds. Dynamics and controls experiments conducted during sea trials in autumn 2007 were focused on demonstrating vehicle control in the longitudinal plane (surge, heave and pitch) and 85

Bowen et al. The Nereus hybrid underwater robotic vehicle

Increasing forward speed

Pitching Flight Level Flight

Vectored Thurst

Foils-Fixed Hover

Fig 7: Nereus’ depth and pitch control modes: lift

dominated low angle of attack flight modes will enable efficient operation at high speeds. Thruster dominated high angle of attack hover modes will produce the climb angles necessary to negotiate steep terrain at low speeds. Foils-fixed hover and the two flight modes were demonstrated successfully in the recent sea trials identifying hydrodynamic parameters critical to maximising vehicle performance during survey and high-efficiency homing descents. 2.9.1. Longitudinal plane control Nereus ’ unusual foil-mounted actuator configuration creates a variety of potential strategies by which the vehicle can control its depth. The efficiency and performance of each is predominantly a function of forward speed, with higher speeds favouring low angle of attack motions to limit drag losses. Fig 7 illustrates Nereus ’ gain scheduled strategy for depth control. Extensive simulation studies (Jakuba et al., 2007) were utilised to develop the following three depth controllers: 1. Foils-fixed hover: hovering for low speed, high angle of attack manoeuvres 2. Zero-pitch flight: level flight for intermediate to high-speed survey at low to medium angles of attack 3. Zero-w flight: zero body-frame vertical velocity (zero-w) pitching flight for efficient high-speed climbs at low angles of attack. The three control modes were demonstrated successfully during the sea trials and tuned for acceptable performance. Automated switching between level flight and foils-fixed hover was demonstrated during a high-speed terrain following experiment at 5m altitude, as depicted in Fig 8. 2.9.2. Terrain following The utility of Nereus as an imaging survey AUV is dependent upon its ability to terrain-follow (maintain a constant altitude above the seafloor). The terrain-following algorithm implemented on Nereus is modelled on the algorithm employed

86

on the successful ABE AUV (Yoerger et al., 1998, 2007). The algorithm attempts to limit control action and provide robustness to noisy altimeter data by maintaining the depth set-point within a pre-specified envelope above the seafloor rather than servoing off altitude directly. In the 2007 sea trials Nereus successfully flew a 5m-altitude photo survey over modest terrain. The result (Fig 8) is noteworthy in that despite the relatively smooth seafloor encountered, two outcroppings were sufficiently steep to require higher climb angles than were expected to be possible in level flight mode. The control program responded to these obstacles by autonomously switching actuator allocation modes from level flight to foils-fixed hover. These results were attained without the benefit of feed forward or integral action in the depth controllers. These features have been implemented and will be engaged in subsequent trials.

2.10. Mission controller This section reports the progress towards developing a mission controller for the high-level control of the Nereus HROV. The mission controller performs the job of the pilot in the absence of a telemetry system. In AUV mode, the mission controller supports fully autonomous survey missions, while in ROV mode, it permits normal pilot-controlled teleoperation of the vehicle yet, in the event of the loss of tether telemetry, will assume control of the vehicle and autonomously complete a preprogrammed mission. The mission controller manages the mission by executing a high-level mission plan and response to events. At present the mission controller is capable of the following: • Condition monitoring: detection and response to vehicle subsystem status, including vehicle depth and Doppler sonar status • Telemetry status monitoring: detection and response to status of fibre-optic Ethernet link • New mission execution: the ability to dynamically load and execute new mission scripts • Simulation mode: fast time simulated missions • Simple command primitives: simple primitives for control of speed, depth, altitude, and heading • Trackline following: trajectory control for survey operations • Abort mission: the ability to perform a controlled mission abort.

2.11. Acoustic telemetry The acoustic telemetry system is designed to send data acoustically between multiple vehicles, including one or more surface ships. The system employs WHOI Micro-Modems and, in the case

Vol 28, No 3, 2009

Nereus AUV Bottom Follower: 5m Altitude During Sea Trials Abroad R/V Kilo Moana, 2007-Nov-26 Veh. Goal Ref. Bot. Env.

Depth (m)

20 22 24 26 28 30 11.75

11.8

11.85

11.9

11.95

12

12.05

12.1

12.15

12.2

12.25

Distance Along Track (km)

Fig 8: Terrain follower performance versus distance travelled during a mock photo survey at 5m altitude:

times when the terrain follower was actively controlling depth are indicated by the grey areas. The algorithm autonomously modulated speed to keep the reference trajectory within a 1m envelope throughout the majority of the survey. This included autonomous switching into foils-fixed hover allocation from level flight allocation when the desired climb angle exceeded that attainable in level flight. Mode switch times are indicated by heavy vertical bars

• An ‘abort’ message indicates that the vehicle should run its abort script. • A ‘ranging ping’ interrogates another modem and returns a travel time between the sender and the receiver. • A ‘cycle init’ message requests data and designates which vehicle is to send and which is to receive.

vehicle Position via Doppler and Acoustic Uplinks 300

data uplink of vehicle position vehicle position

250 200 Y pos [meters]

of the Nereus vehicle, EDO/Straza SP23 acoustic transducers. The acoustic communications program is a Linux daemon that runs on the Nereus control computer. It manages the WHOI Micro-Modem, serving two main roles. First, the acoustic communications process is designed to monitor and report on the status of the modem and keep it properly configured through modem reboots. Second, it is designed to act as a transport layer between the vehicle program and the modem for sending and receiving acoustic data, providing basic format checking for messages sent to the modem and exhaustively logging all acoustic communications-related data. For the recent sea trials, Nereus was equipped with one Straza transducer mounted on the forward starboard hull of the vehicle facing upwards. A second transducer was deployed from of the stern of the ship (the RV Kilo Moana ) facing downwards. The ship transducer was lowered several metres under water and held in place using a bridle that allowed very little side-to-side motion. During vehicle ship communication, the vehicle is designated as the master, initiating all acoustic communications. Messages are sent out in a time division multiple access (TDMA) cycle that is configured at run time. The slave acoustic communications process on the ship responds to acoustic queries as necessary and can send an acoustic abort signal if needed. The following acoustic messages were employed:

150 100 50 0 –50 –100 –50

0

50

100

150 200 250 X pos [meters]

300

350

400

Fig 9: Vehicle trackline overlaid on vehicle

positions transmitted via acoustic uplink • A ‘vehicle status’ message consists of 32 bytes of data containing vehicle status information (position/attitude, velocity and battery charge). • An ‘LBL ping’ interrogates LBL acoustic beacons at up to four distinct frequencies and reports travel times. In the November 2007 sea trials, tests demonstrated the ability to send and receive abort packets as well as respond to data requests with a ‘vehicle status’. This vehicle information proved invaluable during AUV trials, allowing the users to track the progress of the vehicle in real-time during its mission. Fig 9 shows an example of status (position) data received acoustically at the ship in real time and the actual vehicle trackline. The LBL functionality of the acoustic communications process has been fully tested in an LBL transponder net subsequent to this cruise, where

87

Bowen et al. The Nereus hybrid underwater robotic vehicle

the ability of the navigation program to parse Micro-Modem LBL messages and calculate the resulting vehicle position has been demonstrated.

3. Conclusions The combination of AUV and ROV attributes in a single vehicle offers an exciting step forward with great relevance to a range of today’s subsea challenges. In the case of Nereus, necessity has created a certain embodiment of this concept based on a range of innovative technologies, resulting in a relatively inexpensive solution for working at such extreme depths. It should be noted that many of these innovations have potential application at more modest depths. Efficient use of materials, coupled with advances in power storage and lower energy subsystems, leave ongoing challenges in the development of control systems and communications. WHOI will continue to pursue these challenges to best fit autonomous capabilities within an appropriate framework of direct human control over such vehicles.

Acknowledgements In March 2004, WHOI engaged the noted ceramics experts, Drs Jerry and Joan Stachiw of Stachiw Associates, to assist with the development of ceramic pressure housings and flotation for the vehicle. They led the effort to develop, procure and test the ceramic components on Nereus, and provided invaluable technical support through out the programme. We regret that Jerry Stachiw passed away on 25 April 2007. Our heartfelt thanks go to Joan for all of her and Jerry’s contributions to the Nereus project. We gratefully acknowledge the support of the National Science Foundation (NSF) as the lead funding agency, as well as the US Navy Office of Naval Research, the National Oceanic and Atmospheric Administration, the Woods Hole Oceanographic Institution, and the Russell Family Foundation.

References Alameda Jr W. (2002). Seadevil: A totally integrated inertial navigation system (INS) solution. In: Proceedings of the Underwater Intervention Symposium, 27 February–2 March, New Orleans. Aoki T, Tsukioka S, Hattori M, Adachi T, Ietsugu N, Itoh T and Nakae T. (1992). Development of expendable optical fibre cable ROV ‘‘UROV’’. In: Proceedings of IEEE/MTS Oceans ’92, 26–29 October, Newport, RI, Vol 2, 813–818. Bowen A, Yoerger D, Whitcomb LL and Fornari D. (2004). Exploring the deepest depths: preliminary design of a novel light-tethered hybrid ROV for global science in extreme environments. Marine Technology Society Journal 38(2): 92–101.

88

Broad W. (1997). The Universe Below: Discovering the Secrets of the Deep Sea. New York: Simon and Schuster, 432pp. Eustice RM, Whitcomb LL, Singh H and Grund M. (2007). Experimental results in synchronous clock one-way-travel time acoustic navigation for autonomous underwater vehicles. In: Proceedings of IEEE International Conference on Robotics and Automation, 21–27 April, Albuquerque, NM, 4257–4264. Fletcher B, Young C, Buescher J, Whitcomb LL, Bowen A, McCabe R and Yoerger D. (2008). Proof of concept demonstration of the hybrid remotely operated vehicle (HROV) light fibre tether system. In: Proceedings of IEEE/MTS Oceans ’08, 15–18 September, Quebec City, Canada. Freitag L, Grund M, Singh S, Partan J, Koski P and Ball K. (2005). The WHOI micro modem: an acoustic communications and navigation system for multiple platforms. In: Proceedings of IEEE/MTS Oceans ’05, 17–23 September, Washington, DC, Vol. 2, 1086–1092. Fryer P, Becker N, Appelgate B, Martinez F, Edwards M and Fryer G. (2003). Why is the Challenger Deep so deep? Earth and Planetary Science Letters 211(34): 259–269. Gobat J and Grosenbaugh M. (2000). WHOI Cable v2.0: Time domain numerical simulation of moored and towed oceanographic systems. Technical Report WHOI-200008. Woods Hole, MA: WHOI, doi:10.1575/1912/63. Gobat J and Grosenbaugh M. (2001). Application of the generalized method to the time integration of the cable dynamics equations. Computer Methods in Applied Mechanics and Engineering 190: 4817–4829. Howland J, Farr N and Singh H. (2006). Field tests of a new camera/LED strobe system. In: Proceedings of IEEE/MTS Oceans ’06, 18–21 September, Boston, MA, 1–4 (doi: 10.1109/OCEANS.2006.307142). Hunt MM, Marquet WM, Moller DA, Peal KR, Smith WK and Spindell RC. (1974). An acoustic navigation system. Technical Report WHOI-74-6. Woods Hole, MA: WHOI. Jakuba M, Yoerger DR and Whitcomb LL. (2007). Longitudinal control design and performance evaluation for the Nereus 11,000m underwater vehicle. In: Proceedings IEEE/MTS Oceans ’07, 29 September–4 October, Vancouver, Canada, 1–10 (doi: 10.1109/OCEANS. 2007.4449305). Jamieson AJ, Solan M and Fujii T. (2008). Imaging deep-sea life beyond the abyssal zone. Sea Technology 50(3): 41–46. Kaharl V. (1990). Water Baby: The Story of Alvin. New York: Oxford University Press, 356pp. Kinsey JC and Whitcomb LL. (2004). Preliminary field experience with the DVLNAV integrated navigation system for oceanographic submersibles. Control Engineering Practice 12(12): 1541–1548. Larsen MB. (2000). High-performance Doppler inertial navigation experimental results. In: Proceedings IEEE/MTS Oceans 2000, 11–14 September, Providence, RI, Vol. 2, 1449–1456. McEwen R, Thomas H, Weber D and Psota F. (2005). Performance of an AUV navigation system at Arctic latitudes. Journal of Oceanic Engineering 30(2): 443–454. McFarlane J. (1990). ROV-AUV Hybrid for Operating to 38,000 Feet. Marine Technology Society Journal 24(2): 87–90. Murashima T, Aoki T, Nakajoh H, Tsukioka S and Asao Y. (1999). Optical communication system for expendable fibre optics ROV UROV7K system. In: Proceedings of the International Symposium on Offshore and Polar Engineering: ISOPE99, 30 May–4 June, Brest, France, 628–634. National Oceanic and Atmospheric Administration (NOAA). (2000). Discovering Earths Final Frontier: A US Strategy for Ocean Exploration. Washington, DC: NOAA.

Vol 28, No 3, 2009

National Science Foundation (NSF). (1996). The Future of Marine Geology and Geophysics, NSF Workshop Report. Washington, DC: CORE. Newman JB and Stakes D. (1994). Tiburon: Development of an ROV for ocean science research. In: Proceedings of IEEE/MTS Oceans ’94, 13–16 September, Brest, France, 483–488. Piccard J and Dietz R. (1961). Seven Miles Down: The Story of the Bathyscaph Trieste. New York: Putnam, 249pp. Shepard A, Fryer P, Bellingham J, Moore B, Kelly M, Zande J, McCurdy A, Carless J, Ward M and Lemmerman L. (2002). Link Symposium: Sea and Space Experts Join to Develop Undersea Technologies NOAA/NASA Symposium. Kennedy Space Center: NASA. Stachiw JD and Peters D. (2005). Alumina ceramic 10 in flotation spheres for deep submergence ROV/AUV systems. In: Proceedings of MTS/IEEE Oceans ’05, 17–23 September, Washington, DC, Vol 1, 164–171. Stachiw JD, Peters D and McDonald G. (2006). Ceramic external pressure housings for deep sea vehicles. In: Proceedings of IEEE/MTS Oceans ’06, 18–21 September, Boston, MA, 1–7 (doi: 10.1109/OCEANS.2006.306971). Takagawa S. (1995). Advanced technology used in Shinkai 6500 and full ocean depth ROV Kaiko. Marine Technology Society Journal 29(3): 15–25. University-National Oceanographic Laboratory System (UNOLS). (2000). DESCEND Workshop Proceedings, 25–27 September, Arlington, VA. Available from www.unols.org/meetings/1999/199910dcd/descend_ proceedings.pdf, accessed 30 October 2008.

Weston S, Stachiw JD, Merewether R, Olsson M and Jemmott G. (2005). Alumina ceramic 3.6in flotation spheres for 11km ROV/AUV systems. In: Proceedings of IEEE/MTS Oceans ’05, 17–23 September, Washington, DC, Vol 1, 172–177. Whitcomb LL, Kinsey J, Yoerger DR, Taylor C, Bowen A, Walden B and Fornari D. (2003). Navigation upgrades to the National Deep Submergence Facility vehicles D.S.V. Alvin, Jason 2, and the DSL120A. In: Eos Transactions AGU 84(46): Fall Meet. Supplement. Abstract OS32A0225. Whitcomb LL, Yoerger DR and Singh H. (1999). Advances in Doppler-based navigation of underwater robotic vehicles. In: Proceedings of IEEE International Conference on Robotics and Automation 1: 399–406. Yoerger DR, Bradley AM, Walden B, Singh H and Bachmayer R. (1998). Surveying a Subsea Lava Flow Using the Autonomous Benthic Explorer (ABE). International Journal of Systems Science 29(10): 1031–1044. Yoerger DR, Jakuba M, Bradley AM and Bingham B. (2007). Techniques for deep sea near bottom survey using an autonomous underwater vehicle. International Journal of Robotics Research 26(1): 41–54. Young C, Fletcher B, Buescher J, Whitcomb LL, Yoerger DR, Bowen A, McCabe R, Heintz M, Fuhrmann R, Taylor C and Elder R. (2006). Field tests of the hybrid remotely operated vehicle (HROV) light fibre optic tether. In: Proceedings of IEEE/MTS Oceans ’06, 18–21 September, Boston, MA, 1–6 (doi: 10.1109/OCEANS.2006.306819).

89