THE APPLICATION OF A MODULAR AUV TO COASTAL OCEANOGRAPHY: CASE STUDY ON THE OCEAN EXPLORER

Transcription

1 THE APPLICATION OF A MODULAR AUV TO COASTAL OCEANOGRAPHY: CASE STUDY ON THE OCEAN EXPLORER Samuel M. Smith, Stanley E. Dunn, Thomas L. Hopkins, K. Heeb, and T. Pantelakis *Florida Atlantic University Ocean Engineering Dept. 777 Glades Rd. Boca Raton, FL (tel) (fax) (internet) smith@oe.fau.edu Abstract - This paper describes the application of the I Ocean Explorer (OE) series autonomous underwater vehicle (AUV) to a series of coastal measurement tasks. The Ocean Explorer is a small, lowcost, long range AUV whose primary purposes are coastal oceanographic survey missions and support of littoral warfare. Over a dozen different sensor payloads are under development at and for deployment on the OE vehicle. A brief description of the sensors systems and associated design considerations will be presented. The OE A W is a modular vehicle with a dedicated tail or propulsion section that houses the main computation, navigation, and control systems. A forward payloadnosecone section is left free for sensor payloads. Each payload has a dedicated nosecone section built for it so that no physical retrofitting is needed to change payloads. The electrical and control interface is an intelligent distributed control system (IDCS) based on the LONTalk protocol and Neuron chip. All the sensors and actuators on the vehicle are connected on a single serial network. A standard connector with power and network wires permits plug and play reconfiguration. The OE s modular nature provides for more costeffective sensor deployment. and would allow researchers at institutions besides to build payloads for deployment. Interface specification and guidelines will be discussed in order to facilitate this type of cooperation. INTRODUCTION The Ocean Explorer series vehicle is part of a cooperative multi-institutional research effort to develop sensors and platforms for unmanned untethered coastal oceanographic survey. The participants include the University of South Florida, Florida Atlantic University, The University of Hawaii, MIT, NRL Stennis, CSS Panama City, and FIT. The Ocean Explorer (OE) serves a a modular truck for deploying a wide variety of sensor systems. Three vehicles will be built as part of the first phase of the project. Others are contemplated. The major impetus of the project is to devellop technology that more effectively addresses the problems of sampling and survey in coastal oceans. Autonomous underwater vehicles (AUVs) proviide several potential advantages over current methods. By removing the tether, a stable maneuverable platform is possible with 3D sampling paths unaffected by the surface ship s motion. Inclement weather operation is possible. AUVs can fly very close to the sea floor thereby providing continuous wide area coverarge where much biological, chemical, physical, and man made activity takes place. Appropriate sampling paths produce pseudo-synoptic data collection. The absence of a tether makes cost effective simultaneous operation of multiple AUVs possible thereby providing synoptic data. Since AUVs have on board computation and mission control capabilities, adaptive saimpling is possible. Moreover, that fact that AWs are unmanned facilitates military applications such as surveys in hostile coastal waters or for mine counter measures.the advent of reliable underwater acoustic communication will allow cooperative adaptive multiple vehicle sampling. Admittedly, AUVs present many difficult technical challenges especially in navigation, communications, and power. Recent advances on all these fronts, however, make the application of AUVs to coastal oceanography eminently feasible. [ 1,2,3,10,13] This paper will list some of sensor systems and payloads under development as motivation for the design approach taken for the OE AUVs. A description of the OE system is given with design and operational consideration engendered by various payloads. PAYLOADS The different sensors system currently undler development are listed in Table 1. The sheer number of different payloads meant that a conventional 1423

2 approach to AUV construction was not feasible or cost effective. An AW has certain dedicated onboard components that are needed to achieve underwater motion control and navigation independent of the payload. Surface support navigation and tracking equipment may also be needed (Such as an USBL or LBL system and launch and recovery vessel). The nominal cost of the AUV is based on the cost of all these components both on-board and support. One approach to deploying multiple payloads would be to build a separate A W dedicated to each payload. This is cost effective if the payloads all have to be operated simultaneously or if the nominal cost of the AUV is incidental to the cost of the payload. However, for most of the payloads the cost of the AUV/support, is still not an incidental expense. The approach then was to have multiple payloads take turns using one A W platform. This requires that the AUV be modular both in structure and in its electrical, power, sensing, and control systems Some of these systems such as the imaging sonars and buried object detector are in the early development stages and are two big and power hungry to mounted on the A W this year. However the goal is to reduce the size and power requirements so that AUV mounting can be achieved in the near future. A summary of some of the systems that are scheduled to be deployed on the Ocean Explorer vehicles this year is given below. Chirp Side-Scan Subbottom Sonar (CSSB) The subbottom sonar will be used for determining bottom sedimentation layers and surface properties such as volume scattering, acoustic attenuation and impedance. The side scan will provide surface features.this sonar consists of a downward looking Chirp subbottom profiler (2-10 khz) and two side mounted Side-scan transmit and receive arrays (45 khz band center, 10 khz bandwidth). The side-scan arrays also function as receive arrays for the subbottom transducer. The Side-scan is quantitative in that its measurements are repeatable. This due to the fact it uses \high resolution (16 bit) ADCs instead of adaptive gain control. This system was deployed on the Ocean Voyager I1 during May and to our knowledge is the first side-scan subbottom system deployed on a small AUV. Integration into a dedicate OE payload is underway. Basic missions will consist of surveys in an LBL grid off the east coast of Florida and later in the Gulf of Mexico. The AUV can fly at a constant altitude with respect to the bottom thereby reducing power requirements for the sonar. Thc A W also provides a stable platform. The vehicle attitude sensor outputs are feed to the CSSB to compensate for any residual motion. The primary operational constraint engendered by this payload is that the AUV fly level with a pitch and roll angle less than 5 degrees. This ensures that a significant portion of the sonar signal is returned toward the AUV as determined by the beam width of the sonars. Self-Motion Detection. Little data has been collected on the self-motion characteristics of small AUVs operating in shallow water. This is of concern for synthetic apeture sonars which are highly sensitive to vehicle motion and to some extent a concern for other types of imaging sonars. The Ocean Explorer is equipped with a lowcost AHRS system consisting of a 3 axis magnetome- Description Nutrient nitrogen & ammonia Trace element distribution, Mn, Fe,HS, PC02, PH Microparticle analysis Fluorescence detection of dissolved organic matter Water column particulates Biomass quantification Small-scale turbulence Sea Bottom Characterization Surface & subbottom features & sediment properties Buried Object Detection Bottom classification & albedo package Acoustic Imaging High resolution 3-D imager High resolution 3-D imager Sensor Fiber-optic spectrophotometer Fiber-optic spectrophotometer Fiber-optic spectrophotometer Laser fluorometer Optical particle counter Acoustic particle counter Shear probes & electromagnetic current sensor Chirp sidescan subbottom sonar Bottom penetrating sonar Hyperspectral imager, radiometers Fourier transform beamformer Hilbert transform beamformer Institution, Ambient acoustic imager Phased array correla- ' tion sonar Navigation & Communicafion Acoustic Modem Vehicle self motion Long baseline sonar AHRS, Accelerometers, DVL RLG INS. CSS, MIT 1424

3 ter and solid state gyros that measure roll, pitch, and yaw rates. Accelerometers measure heave, sway, and surge accelerations. These will be used to measure the self-motion of the vehicle at various depths and at various sea states. By correlating the self-motion data to operating conditions, some measure will be gained of the limitations of small AUVs as sensor platforms for motion sensitive sonar systems such as synthetic aperture and imaging sonars. A side-by-side comparison of the low cost system with a ring laser gyro system under development at NCSS at Panama City will be conducted. One extension of this study would be to determine control strategies that minimize vehicle self motion. Small Scale lhrbulence A small scale turbulence package under development at will be used to measure local turbulence properties. These can be used to calibrate dispersion and suspension rates of sediments and nutrients and acoustical propagation characteristics. The sensor package consists of shear probes and an electromagnetic velocity sensor mounted on the nose of the vehicle. An accelerometer bank mounted in the nose will be used to subtract out vehicle vibrations. This package will require long straight transects with a minimum of vehicle motion. CTD A standard component on the Ocean Explorer vehicles is a Falmouth micro CTD. The OE uses the pressure gage on the CTD to determine depth. Any time the OE is in operation it will collect CTD data as a by-product of whatever else it is doing. Possible missions includes mapping inlet plumes and gulf stream eddies. Acoustic Modem in conjunction with EG&G has developed an acoustic modem designed for high reliability in doppler and multipath environments at data rates of around 2k bps. This modem uses narrow band phase encoded pulses each centered on a sub frequency band. Several sub frequency bands are used in parallel to increase the data rate (MFSK). The modem uses a TIDSP320C30 DSP chip. A low speed acoustic release type modem is used to wake up the high speed modem. The modem will provide real time telemetry, mission control updates, and the possibility of coordinated multiple vehicle operations. OE AUV DESIGN CONSIDERATIONS Currently AUV are not used as continuous duty platforms. Most of the reported missions are of the technical demonstration variety or if for Oceiinographic purposes more of one shot type missions [7]. A basic design goal for the Ocean Explorer was daily operation. A practical scenario for daily operation would be limited to daylight use with no more than one launch and recovery cycle per day. This means a minimum of 12 hour mission duration with time for mission preparation, launch, recovery, and data analysis taking up the balance of a 16 hour day. Battery charge time should be 8 hours or less. Furthermore the battery technology should be robust and inexpensive enough to support daily use. The targeted missions are in shallow coastal waters along the Florida coast and in the Gulf of Mexico. Consequently, the designed maximum operating depth for the Ocean Explorer is 300 meters. As a general rule the AUV range is maximlzed when the power requirements for the hotel load and propulsion are approximately the same [4]( See Appendix). The OVII for example has a hotel load of around 50 watts. The AW should be small enough to be handled off from a small boat yet large enough to hold most sensor systems. Furthermore, the speed of the vehicle is limited by the response times of certain sensor systems such as a CTD.These factors indicate that a practical design compromise results in a vehicle with a low drag hull, between 6-10 ft. long, less than 2 ft. in diameter, and running at about 3 knots. Another consideration limiting the practical continuous use of the AUV is the down time required to reconfigure the vehicle for different mission payloads. As discussed in the previous section, because the cost of the AUV s dedicated components cost is not imcidental to the payload s cost it is not cost effective to devote an AW to each payload. The ease of reconfiguring the AUV for different payloads significantly impacts its ability to sustain daily operations. In order to maximize usability, field reconfiguration is necessary, that is swapping payloads while at sea. In large vehicles payload areas can be relatively spacious and it becomes possible to have a standard payload configuration such as a 21 inch parallel midbody. In contrast, small vehicles such as the Odyssey or Ocean Voyager I1 have flooded fairings where space is shared between dedicated vehicle components and payload. Changing the payload often involves custom mounting and rearrangement of internal components such as buoyancy foam etc. Depending on the nature of the control system, electrical wiring and vehicle software may also have to be redone to accommodate changes. This reconfiguration process can take days or weeks to accomplish thus limiting the multi-payload usefulness of the vehicle. This was a major motivation for the Ocean Explorer 1425

4 design modularity. Structure The Ocean Explorer consists of a tail section propulsion, navigation, and control section and modular nosecone/payload sections. The hull shape is a modified hydrodynamically efficient Gertler body shape with a maximum OD of 21 inches and a length of 7 ft. The hull is a flooded fiberglass fairing with individual pressure vessels to house the electronics. The OE can accommodate 21 inch diameter parallel mid-body inserts of 1, 2, or 3 feet long. The parallel mid-bodies only increase drag by about 4% over a true Gertler body. The simplicity of only having to build one nosecone mold and the modularity of the parallel midbody setup outweighed the marginal increase in drag. Aft mounted rudder and sternplanes provide directional and depth control. To accommodate the changes in control and stability associated with the different hull sizes, matching sets of control surfaces have been designed that can be easily swapped. The tail section is about 4 feet long and houses the thruster, propeller, rudder, sternplane, servos, batteries, DVL, Depth sensor, and Main computer. The nosecone is about 3 feet long and is left open for payloads except for an emergency drop weight system. Much effort went into reducing the size of the servo, thruster, and gear box units to minimize the size of the tail section. A simple bayonet type mounting ring connects the tail and payloadhosecone sections. This means that each payload can have a dedicated nosecone section so that no physical retrofitting is needed to change payloads, The propulsion system consists of a brushless DC motor with a 5-to-1 reduction gear driving a 3 bladed high efficiency propeller spinning at rpm. The propeller s projected efficiency is a least 85%. In addition, losses in the motor controller, motor, gearbox and seals bring the projected power train efficiency to around 50% [6]. This is comparable to the measured values of the OVII [5]. A plot of the power requirements for the different vehicle lengths at various speeds is given in Figure 1 below. The projected hotel load for the Ocean Explorer is between 50 and 60 Watts. The equivalent propulsion loads occur around 3 knots. For the longest vehicle (10 ft.) the propulsion load at 3 knots is about 70 watts. This and a conservative hotel load of 60 Watts means that for 12 hour operation at least 1560 Watt-hours of energy are needed. Several battery technologies were examined. A compromise between storage density, cost, modularity, and reliability resulted in the selection of Ni- Cad D cells for the power system. To maximize 300 n 250 v) cn + g ft. I Speed (knots) Figure 1: Propulsion power requirements for Ocean Explorer series AUVs usability and avoid memory effects an intelligent battery monitoring and charging system was developed. The final system consists of 8 modular battery packs, each with 42 D cells and a dedicated microprocessor based monitoring and control system. Each pack supplies 4.75 Amp-hours at a nominal 50 volts at a 5.7 Amp discharge rate. The total energy storage then is 1900 Watt-hours. A single powedsignal cable connects the battery packs. Each pack has programmable switches that allow the each pack to be connected or disconnected to the power bus. The packs are discharged in a round robin fashion. The modularity of this arrangement means that adding battery packs in the payload section is easily done. Each battery pack is housed in a 4 diameter by 18 long aluminum cylinder. The main computer is housed in a 10 diameter by 20 long aluminum pressure vessel. The battery cans surround the main PV in a honeycomb arrangement. A diagram of the various vehicle configurations is given in Figure 2 below. Pictures of the actual vehicle are shown in Figure 3. Intelligent Distributed Control System. Ease of reconfiguration requires not only structural modularity but electrical and control as well. To this end an intelligent distributed control system (IDCS) architecture was used. This was an extension of the experience gained with this type of approach on the OVII. The IDCS is based on the LONTalk protocol and Neuron chip. AI1 the sensors and actuators on the vehicle are connected on a single serial network. A standard connector with power and network wires is daisy-chained to each of the pressure vessels and devices. Adding devices to the vehicle consists of plugging into this network. Software rewrites are min- 1426

5

6 1428 (c) Figure 3: Occan Explorer. (a) Hull (b) Nosecollc mounling (c) Tail Section

7 ~ imized because network communications is at a high level. The IDCS is also used to implement health monitors in each of the pressure vessels and the intelligent battery system. The IDCS approach also facilitates upgrading of system components or adding functionality [8,9,11]. The conventional approach to AUV control hardware is to have a central computer system with analog, digital, and serial point-to-point connections between a bank of VO boards in the main computer and the sensors and actuator systems. This means that each sensdactuator requires its own wire(s). This requires many cables, pressure vessel penetrations, and isolation circuits. Reconfiguring the vehicle requires rewiring the I/O and reprogramming the low level device drivers reading in the U0 lines. In the distributed control system approach, the sensors and actuators are connected on a serial bus and communicate with a networlung protocol similar to a Local Area Network. Typically a microprocessor or communications chip is used to interface between the network and the individual sensor or actuator. Each connection to the network is called a node. Information is sent over the network as messages. The LONTalk/Neuron chip set provide very low cost per node networking with a state of the art protocol. This means that only one cable is needed to connect nearly all the systems on the vehicle. The only limitation is bandwidth. The LONTalk network on the Ocean Explorer runs at 1.25 Mbps. This capacity can be upgraded by adding routers and partitioning the network. This is not fast enough for real time video or sonar. Since without a tether, however, there is no communications path to the surface that can sustain real time video or sonar data rates, this type of data is dumped to a storage system to be retrieved later. Adding functionality to a node (such as adding additional sensors or actions) does not require rewiring only adding additional messages to the software. Moreover, the physical characteristics of the sensor could change (such as changing manufacturer) but the network message data wouldn t. For example each intelligent battery pack has 16 ADCs inputs that mon base T Ethernet \ 1 1 Ocean Explorer Distributed Control Network VxWorks 8 MegDril NavigationlAutopi lot RF Ethemet Gateway kbps RF Pressure Vessels -0 Pressure Vessels L--.---: i Pressure Vessels Figure 4: Intelligent distributed control system network layout for the Ocean Explorer AUV. 1429

8 itor voltage and current of the 7 battery stacks plus temperature and humidity. In a conventional setup, this would require 128 analog wires running to the main computer. Adding additional batteries would only exacerbate the problem. With the IDCS approach the ADC information is collected at each node and then transmitted as network messages. Changing the number of battery packs does not require rewiring the main computer. The advantages of the IDCS approach for the payload designer and user are significant. Since nearly all the onboard sensor data is transmitted over the network, this information is available for use by the payload without special programming of the main computer system. In other words there is no overhead associated with gaining access to the data. Say for example that a payload would benefit from knowing the vehicle's pitch and roll. If the payload had a LON- Talk node it could retrieve that information off the network without having a special connection to the vehicle's main computer. It would also be possible if desired and if permission was given for a payload to control of some of the subsystems on the vehicle. For example the emergency drop weight could respond to a leak in a payload pressure vessel. The IDCS approach allows for modularity in information usage and payload configuration. The are a number of ways of interfacing a payload's electronics/computer to a Neuron Microprocessor. One way is through an RS-232 serial port. Another way is using a synchronous peripheral interface or digital U0 lines on the Neuron. The IDCS for the Ocean Explorer is shown in Figure 4 below. The LONTalk network is used for sending information between nodes. The nodes on the network can perform low level control functions. High level software and software that is computationally intensive is performed on a VME VxWorks based main computer system. The main computer interfaces to the LON with a VME/LON card. The main computer has a ethernet based TCP/IP link for up loading mission plans and downloading data. OCEANOGRAPHIC TOOL The primary advantages of AUVs as oceanographic sampling platforms consist of the ability to sample at a variety of temporal and spatial scales throughout the water column relatively unaffected by surface conditions. On board computation systems makes adaptive sampling possible. This is where the vehicle changes its path based on sensor data to optimize its sampling strategy, such as gradient following for source localization. Because AWs are untethered synoptic data can be obtained through simultaneous operation of multiple AUVs. As a case study, this section will discuss the navigation and sampling strategies possible for the Ocean Explorer. The basic sensor suite on the OE includes a pressure sensor, attitude and heading reference system, doppler velocity log(dvl), differential global positioning system (DGPS). These provide depth (< 3 cm), heading (< OS"), roll (< lo), pitch (lo), ground speed (< 0.2 cds), water speed ( < 1 cds), altitude above bottom (1% fs), lat long on surface (< 3m). Navigation The Ocean Explorer has an autopilot running on the main computer that controls to commanded depth or altitude, heading, and speed setpoints [12]. Missions consist of a set of waypoints of desired XY locations, depthdaltitudes, and speeds. The minimum altitude for save control of the vehicle is about 2 meters. A position estimator modille determines the vehicles horizontal position by whatever navigation sensors are available. Based on this estimate a seeker module generates setpoints for the autopilot that will bring the vehicle to each waypoint in turn. The following navigation modes can be employed by the vehicle. Dead Reckoning - The DVL determines ground speed and the compass determines heading. Position accuracies of k 10 m per km traveled are obtainable. Periodic surfacing with DGPS updates can bound the positional error. The limitations are the maximum ground lock altitude 30m and the loss in effective sampling time when getting DGPS updates. If ground lock is not obtained then the DVL provides water speed. Current velocities could be estimated given DGPS updates. Long Baseline (LBL}- Preinstrumenting a site with an LBL transponder net allows for continuous underwater sampling with about 5 meter accuracy. The limitations is the size of the LBL net (1-2 km square in shallow water) and the need to preinstrument the site. USBL Trucking - An ultra short baseline (USBL) system such as the ORE Trackpoint can be mounted on a surface ship with a transponder on the AUV and a track of the AW maintained. Using an acoustic modem updates of the AW position as determined by the USBL can be given. This would obviate the need for DGPS updates, The USBL has position accuracies of < 10 meters out to ranges of 2 km. For many oceanographic missions it is not so important that the AW no where it is but where it was when a sample was taken. Thus even without an acoustic modem the USBL could be used to back calculate the AWs positions after the mission was over and position stamp 1430

9 the data. Sampling Strategies The maximum nominal system sample rate for controllers and sensor systems on the Ocean Explorer is 8 Hz. Many sensors sample slower than this rate. At a speed of 3 knots (1.54 ds), s'amples could be spaced as closely as 20 cm. For a 12 hour mission a total of 55 km of lineal distance could be covered by the AUV. The spatial distribution depends on the sampling path with high spatial resolution along the vehicle track. Take for example a bottom chsification task where the vehicle follows bottom at a fixed altitude. If an LBL navigation scheme is used the AUV could stay under the whole day. The area to be surveyed is a rectangular patch of the sea floor. One way to do this is with a "lawnmower" pattern of evenly spaced transects along the bottom (See Figure 4). Given the range of the vehicle r, width of the pixtch w, and length 1 along which the transects parallel, the spacing of transects is given by, s=wl r-w Thus for a 55 km range and a 1000 x 750 m rectangular patch, the transects could be spaced about 15 meters apart.alternatively if only a 100 m spacing is needed then area that could be surveyed with w given by w = (rs)/(l+s) isabout5xlkm. Alternatively consider an example where the AUV has to surface periodically to get a DGPS fix. The useful time spent near the bottom and hence the S --ib+ 1 Y- Figure 5: Lawnmower survey pattern. effective coverage is reduces. Suppose the AUV has to surface for a DGPS fix every 2 transects (1 km transects) to bound its positional accuracy while operating at 20 meter depths. The ascent/descent slope for a nominal 20 O pitch is about 3. That is the vehicle has to move 3 meters for each meter it ascendsldescends. An optimistic time to acquire a DGPS fix is about a minute this adds 90 mleters of travel on the surface. Thus during a surfacing cycle the AUV would travel about 210 meters.this translates roughly into a 10% reduction is effective range. This assumes that the AUV surfaces during one of the cross track legs connecting the transects. The effectiveness ratio can be approximated by, e = I-- 2da + g where d is the average depth, a the ascent slope, g the distance spent on the surface, and U the distance underwater between fixes. U CONCLUSIONS AUV technology lhas been in the research and development phase for some time. The Ocean Explorer represents a maturation of that technology to the point of a continuous duty oceanographic tool. Over the next year the AJSF team will gain valuable experience in this pursuit. The number of sensor systems adapted to AUV use will increase tremendously. The OE's modular nature would allow other researchers at institutions besides PAU to build payloads for deployment on the OE. Diagrams and guidelines of the physical interface (quick mount) and the power/ LONTalk cable can be obtained from. Because of the sensor and actuators are on the network a payload could supply its own high level control system. It should be possible for outside scientists to package their sensors in a payload section to be deployed on one of the Ocean Explorer vehicles. REFERENCES Bellingham, J. G., "Capabilities of Autonomous Underwater Vehicles," Scientific Environment Data Collection in Autonomous Underwater Vehicles, ed J. Moore, MIT Sea Grant Report 92-2, Bellingham, J.G., Goudey, C.A. Consi, R.R. and Chryssostomidis, C., A Small, Long- Range Autonomous Vehicle for Deep Ocean Exploration, Proceedings of the Second International Offshore and Polar Engineering Conference, San Francisco CA pp ,1992. Blidberg, R. and G. Sedor, "Report on an Interdisciplinary Workshop to Assess the Scientific Needs for a Long Range Autonomous Underwater Vehicles," U. of New Hampshire,

10 Bradley A. M., Low Power Navigation and Control For Long Range Autonomous Underwater Vehicles, Proceedings of the Second International Offshore and Polar Engineering Conference, San Francisco CA pp , Heeb, Karl, Ocean Voyager 11: Drag Analysis and Propulsion System Efficiency Calculations. Dept. of Ocean Engineering Technical Report # 2S002R00,1994. Heeb, Karl, Ocean Explorer Series: Envelope Description, Propulsor Comparison, Amphour Calculations, Dept. of Ocean Engineering Technical Report 3M001R00,1994. MSEL, An Overview of Autonomous Underwater Vehicle Technology for Ocean Scientists; MSEL Report No , NSF Grant OCE , Marine Systems Engineering Lab, U. of New Hampshire, Durham NH, June MC Neuron Chip Distributed Communications and Control Processors, Motorola 1993 Neuron C Programmers s Guide, Echelon 1992 Smith S.M., and Dunn, S.E. The Ocean Voyager 11: An AUV Designed for Coastal Oceanography. AUV 94, Cambridge MA *S.M. Smith, Implications of Low-Cost Distributed Control Systems in UUV Design, AUVS Conference, Detroit MI, May S.M. Smith, G.J.S. Rae, D.T. Anderson, and A.M. Shein, Fuzzy Logic Control of an Autonomous Underwater Vehicle, Control Engineering Practice, V012. NO. 2, pp ,1994 S.E. Dunn, S.M. Smith, P. Betzer, T. Hopkins, Design of Autonomous Underwater Vehicles for Coastal Oceanography, Underwater Robotic Vehicles: Design and Control, J. Yuh Editor, TSI Press, 1994, pp , book chapter. 2 zero gets h = ks. That is maximum range is obtained when the hotel and propulsion load are equal. APPENDIX The rule of thumb for determining the maximum range for a fixed battery capacity is based on the simplifying assumption that the propulsion power required is proportional to the speed squared. This neglects changes in efficiency of the power train at different rpms and changes in battery capacity due to different discharge rates. The approximate range of the vehicle is given by, cs range = - h + ks2 (3) where C is the battery capacity in Watt-hours, s is the vehicle speed, h is the hotel load in Watts, and k is a proportionality constant. Solving (3) for iis maximum by differentiating with respect to s and setting equal to