Вы находитесь на странице: 1из 5

PrOceedingS of the 2003 IEEBRSJ

Inn. Conference w1 intelligent Robots and Systems Las Veaas. Nevada October 2003

Design of a Prototype Miniature Autonomous Underwater Vehicle


A d i t y a S. Gadre Jared J. Mach Daniel J. Stilwell The Bradley Department of Electrical and Computer Engineering Virginia Polytechnic Institute and State University Blacksburg, VA 24061 {agadre, jmach, stilwell}@ut.edu,
Abstmct- Platoons of cooperating autonomous underwater vehicles have the potential to contribute significantly to scientific investigations in the marine environment. Platoons of vehicles can survey large areas, adaptively track and measure timevarying processes such as tidal fronts and algal blooms, and they are robust to single-point failures. We have developed a prototype miniature low-cost autonomous underwater vehicle t o address the platform requirements of these missions. The vehicle is designed as a test-bed for the development of distributed control and estimation algorithms, and for experiments in advanced navigation and control.

Carl E. Wick
Systems Engineering Department United States Naval Academy Annapolis, MD 21402 wicL&hsn,a. edu

I. INTRODUCTION
Platoons of cooperating autonomous vehicles have significant and well-know advantages over single vehicles operating independently. Cooperating vehicles can rapidly survey large areas, they can survey in four dimensions (space and time), and they are robust to single point failures. It is also hypothesized that the individual vehicles in a platoon of cooperating vehicles could be less expensive and less sophisticated than vehicles designed to operate independently. Most research on multiple cooperating vehicles is conducted using terrestrial vehicles. Yet the marine and ocean science communities would benefit significantly from successful use of cooperating underwater vehicles. Potential applications include surveying chemical plumes, mapping oceanographic processes (e.g., tidal front), and mapping biological populations. For underwater vehicle applications, the complexity and cost of deployments with multiple underwater vehicles is a major factor limiting experimental activities. To facilitate experiments with platoons of coop erating autonomous underwater vehicles (AUVs), we have embarked on a program to develop a miniature low-cost AUV that operates robustly in the field. Our prototype miniature AUV is described herein. Other small AUVs, such as REMUS [I] and Ranger (1 2 could be utilized for multi-vehicle research, but their costs and/or infrastructure requirements are prohibitive. At a cost of less than $3,000 in parts, the prototype AUV described herein is dramatically less expensive than currently
D. J. Stilweli gratefully acknowledges the support of the National Science Foundation via grant IIS0238092, Office of Naval Research via grant N00014-03-1-0444, and the Woodrow W. Everett, Jr. SCEEE Development Fund in cooperation with the Southeastern Consortium of Electrical Engineering Mueatom.

available AUVs. The challenges of acquiring and deploying a platoon of AUVs are the principle motivation for developing the AUV described herein. These challenges are exemplified by the very limited literature of reported experiments with platoons of underwater vehicles. A notable exception was described in (31. Leader-follower control was demonstrated using Odyssey vehicles with a modified ultra-short baseline transponder on the lead vehicle. Though the control algorithm for the follower vehicles was somewhat ad hoc, the system displayed significant robustness. In 141, the development of a new autonomous underwater vehicle designed for laboratory experiments with platoons of vehicles is described. Field operations where multiple AUVs operate in the same area, but are otherwise uncoordinated, have also been reported [Si,161.
. ,.. ,

....
..
'

-.
j

... .

Fig. 1. filly assembled A W .

Fig. 2. Hull and nose removed.

11. A

MlNiATURE

AUV

The principal goals of our miniature AUV design effort is to produce a vehicle that is small, inexpensive, easy t o deploy, and yet fully functional as a platform for marine

survey missions. The current status of our efforts to meet this goal is represented by the miniature AUV described herein, known as MODO. The design of MODO is based on experience with USNA1, a small AUV built to demonstrate that the basic components of an AUV can he packaged in a 3 inch diameter tube (71, [SI. USNA-1 has limited utility as a field-deployable AUV. It has no mission sensors and limited navigation capabilities. It can navigate only via dead reckoning using a gyro-stabilized compass and a depth sensor. MODO builds upon USNA-1 with significantly enhanced propulsion, sensing, navigation, and communication capabilities. Overall specifications for MODO are listed in Table I. Sensor payload is described in Table 11. The design considerations and goals for MODO are derived from anticipated mission requirements as well as the goal of having a small, low-cost AUV that operates robustly in the field. The overall design objectives are: (1) minimize cost by using inexpensive materials and simple mechanisms that are easily fabricated, (2) minimize size (mostly achieved by in-house printed circuit hoard design), (3) maximize range and endurance by choosing a streamlined shape and maximizing energy storage density, (4) address modularity, (5) achieve 200m depth. The miniature AUV must also satisfy the requirements of near term experimental objectives. These include: 1. Using multiple AUVs, sample dissolved oxygen in a large (-100km') coastal lagoon over a 1-2 hour period. 2. Using multiple AUVs, map and track tidally driven gradients in and near coastal river outflows. MODO addresses all of the intended design objectives and costs less that $2,000 in components. Modest upgrades of MODO that are required to more fully meet ow objective of a robust field deployable AUV are discussed in Section 1 1 The next generation field-deployable vehicle costs less 1. that $3,000 in components.

Fig. 3. Tail Assembly and Rear of Payload R m e .

that no ball bearings are used to support the stainless steel fin shafts in their aluminum housings. The fins are made from a 2-part plastic resin and are cast from a rubber mold. The payload frame, shown in Figure 2, consists of two pieces of aluminum bar stock that are rigidly mounted t o the tail section. The forward bulkhead, shown in Figure 4, is an aluminum disc machined to accept an o-ring bore seal. Penetrations through this bulkhead include a pressure transducer (depth measurement) and an 8-conductor electrical connector. A removable handle allows the user to install and remove the bulkhead.

A. Mechanical Description
The mechanical structure of the AUV consists of a tail section, rails for mounting electronic components, a forward bulkhead, hull, and nose. Overall dimensions are given in Table I. The tail, shown in Figure 3, is made up of high-density foam bonded to a section of aluminum tubing. High density foam is chosen for our prototype work since it is easy to machine and inexpensive. We anticipate using PVC for the final version of the AUV. After machining, the foam is sealed with an expoxy resin. The aluminum tubing is machined to accept an o-ring bore seal and slots for three servo motors that actuate the tail fins. The propeller driveshaft has a double o-ring seal. The shaft is machined out of stainless steel and has lubrication-free ball bearings on the inside and outside of the seals keep the shaft aligned with atuminum housing. The driveshaft is connected to the drive motor by means of couplers on the drive motor and shaft and a 'dogbone' shaft to connect the couplers. Three independently actuated fins, spaced 120' apart, use shaft seal designs similar to the driveshaft seal, except

Fig. 4. Front of Payload Frame and Front Bulkhead

Standard 3 inch Schedule 40 PVC pipe makes up the AUV hull. Expansion of the vehicle payload is accomplished by simply extending the length of the payload frame rails and extending the length of the hull. The nose cone is machined from a solid piece of PVC. A pad is machined into the top of the cone for mounting the GPS receiver

843

antenna. The inside of the nose cone is hollowed out to allow for mounting of sensors that require exposure to water, such as the sonar altimeter. The nose attaches to the hull via three dowel pins.

has six built-in serial ports, and a custom engineered printed circuit board was designed to provide 5 additional serial ports. One of the Rabbit's serial ports is multiplexed to provide 1 x 4 serial port expansion using line drivers and buffers. A dual UART provides two additional IjOmapped serial ports: making total number of serial ports 11. Out of these, two serial ports are Rs232 while all others are TTL level. The propeller is driven with a 24V, 40 watt, brushless DC motor. A custom-engineered printed circuit board (PCB) was designed to drive and control the brushless motor along with the three servo motors. It performs electronic commutation of the brushless motor and controls motor v e locity using a quadrature encoder signal from the motor and pulse-width modulation of the motor phase voltages. The motor controller communicates directly with the Rabbit microprocessor over a serial link. The motor controller sources up to 3 amps at 50 V.
C. Sensors

Fig. 5 . Hull and Nare Cane

E. Electronics
Power is provided by a battery stack consisting of 4/3 A size NiMH cells. The battery stack provides 80 watthours at 21.6V. Regulated power at 3.3V, 5V, and 12V is provided by a miniature switching power regulators (Power Trends 7 8 S b H C ) . A separate 3V lithium cell powers the GPS receiver's real-time clock and allows it to save almanac data. A small mechanical relay is used to open and close the main battery circuit through a simple shunt switch in a water tight connector in the forward bulkhead. The AUV control system is centered around a Rabbit 3000 &bit microcontroller (RCM3200 module), running at 44.2 MHz, and programmed in C. The microcontroller interfaces directly with a custom-engineered serial port expansion card and external 8MB FLASH memory. An u p grade path to a laptop HD for high density data storage exists. A block diagram showing the data interfaces with the microcontroller is shown in Figure 6.

Fig. 6. Control System.

All data communication within the AUV is implemented using serial communication. The Rabbit microcontroller

A custom-engineered printed circuit board enables processing of analog sensor signals, including depth transducer, accelerometers, and rategyros. Three-axis acceleration is measured by MEMS accelerometers (ADXLlOS), and 3-axis angular rate is measured by miniature rate gyros (Gyration Micrc-Gyro 100). A network of operational amplifiers provides signal conditioning, including scaling and shifting. Analog-to-digital conversion is accomplished with an 8-channel 12-bit serial ADC (Maxim MAX1270). A PIC microcontroller manages the ADC conversion process, provides digital filtering, and interfaces to the Rabbit microcontroller via the serial bus. The PIC microcontroller samples all analog signals at 2 KHz, averages the samples, and provides them to the Rabbit microcontroller at 10 Hz. Fast sampling and averaging reduces noise by more than an order of magnitude. An off-the-shelf electronic compass (PNI TCM2-20) provides magnetic heading, pitch, roll, and internal temperature. Pitch and roll measurements are also provided by accelerometers on the analog sensors card that measure the gravity vector with respect to body fixed coordinates. A GPS receiver (Motorola M12 OnCore) provides precise position when the vehicle is on the surface. The GPS receiver can provide data in NMEA 0183 data format (at 4800 baud) or Motorola binary format (at 9600 baud). Since hinary data provides more information faster, it is used instead of NMEA format. Time-to-first-fix is reduced by synchronizing time with the Rabbit Microcontroller, which in turn receives time from the host PC. Use of a GPS receiver for precise navigation implies that the AUV will periodically surface for a GPS fix. This is a reasonable assumption given the shallow water applications for which the AUV is intended. Yet an important consideration is the time-to-first& after the AUV has been submerged and returns t o the surface. If time-to-first-fix is too long, then the AUV will spend an excessive amount of time on the surface waiting for a GPS &. To determine time-tofirst.& after the AUV has been submerged, we submerged
644

the GPS antenna, brought it, to the surface after 5 or 10 minutes, and recorded the time-to-first-fix. Histograms of the results of these experiments are shown in Figures 7 and 8. Using a simple Student4 distribution, we determine that if the AUV has been submerged for 5 minutes, then time-to-fist-& occurs within 5 *0.61 seconds 90% of the time, and that if the AUV has been submerged for 10 minutes, then time-to-first-fix occurs within 14.67 i 4.01 seconds 90% of the time. Of course, the applicability of our results is limited since only a small sample of data was taken on a single day. These initial experiments suggest that GPS is a good source of precise navigation data for shallow-water missions where it is reasonable for the AUV to periodically surface. The AUV will not need to spend an excessive amount of time on the surface waiting for a GPS fix.

material. The transducer housing, shown in Figure 9, also contains the transmitter. The receiver is approximately the same size as the transmitter and is housed inside the AUV.

Fig. 9. Custom-engineered sonar altimeter.

6 7 8 9 1011121314 time 10 fix (secs)

Data communication between AUV and the host PC is standard Rs232 serial communication over a Freewave 9OOMHz radio modem. The modem operates at 57600 baud full-duplex (115Khaud total). It implements data encoding and error checking internally, freeing time consuming tasks from the Rabbit microcontroller.

111. NEXTGENERATION MINIATURE AUV


Fig. 7. Time-to-first-fix after being submerged for 5 minut-

A next-generation AUV has been designed based on lessons learned from the design, fabrication, and initial testing of the prototype AUV MODO. The next-generation AUV retains most of the features of MODO while incorprating a few key modifications. A . Mission Sensor Location
Mission sensors for the next generation AUV are mounted on the exterior of the AUV and the nose is not flooded. In contrast, mission sensor transducers are placed inside the flooded nose of the prototype AUV MODO, while associated electronics are placed inside a water tight bulkhead immediately behind the nose. By moving mission sensors to the exterior and removing the forward bulkhead, the overall length of the AUV is reduced, and significant flexibility in modifying the mission sensor payload is obtained. Of course, the costs of mounting mission sensors externally include reduced hydrodynamic efficiency. Figure 10 shows the next generation AUV with a YSI 600XL CTDlDO sonde.

0
1
3

9 1 1 13 15 17 19 21 23 25 21 29

time to fix (secs)

Fig. 8. Tim-to-first-fix &er being submerged for 10 minutes

TWO custom-engineered miniature sonar altimeters are mounted on the AUV. One faces down to determine ve? hicle altitude while the other faces forward for obstacle l' InCEased Diameter avoidance. The altimeter operates at 450 kHz and has a The prototype A W MODO requires 635 grams of lead resolution of approximately 2 cm between 8cm and 9OOcm. ballast, restricting the weight of components that will be Due to a large transducer surface relative to the operat- added in the future. The next generation vehicle is 15 cm ing frequency, the beamwidth is approximately 5 degrees. shorter than MODO, and thus its displacement is signifThe transducer consists of a small sheet of piezo-electric icantly reduced. To achieve an additional safety margin

845

hide for complex 3-D sensing, in Proceedings of the IEEE/MTS OCEANS, pp. 2043-2045, 2001. 1 1 H. Singh, J . Catipovic, R. Eastwood, L. Freitag, H. Henricksen, 3 F. Hover, D. Yaerger, J. Bellingham, and B. Moran, An integrated approach to multiple AUV communications, navigation, and docking, in Proceedings of IEEE/MTS Oceans, (Fort Laud. erdale, Florida), pp. 59-64, September 1996. 1 1 R. Bachmeyer and N. Leonard, A test-bed for multi AUV exper4 iments, in Proceedings of the the 1Oh International Symposium on Unmanned Untethered Submersible Technology, (Durham, New Hampshire), August 2001.

Parameter Length ~ i Mass

CPU
Power

Specification 76.0 cm 9.6 ~ cm ~ t ~ 4.5 kg (without CTD/DO sonde) DIMM-PC (133 MHz AMD Elan SC520) 22.2 volt, 142 watt-hour lithium polymer battery stack

Вам также может понравиться