International Journal of Engineering & Technology IJET-IJENS Vol:10 No:02

81

Development of a Vertical Self-balancing Experimental Autonomous Underwater Vehicle Kamarudin B. Shehabuddeen and Fakhruldin B. Mohd Hashim Department of Mechanical Engineering Universiti Teknologi PETRONAS, Bandar Seri Iskandar, 31750 Perak Darul Ridzuan, Malaysia Tel: +6-05-368 7030, Fax: +6-05-3656461, E-mail: [email protected] or [email protected] Abstract–––Autonomous Underwater Vehicles (AUVs) are operator and tethered umbilical free. Ground motion as a result of seismic activity could result in underwater slope instability and land slide. Other geo-hazard issue such as shallow gas hydrates poses threats to human diver. Therefore, after the process of several design concepts generation and evaluation, the concept of revolutionary human-like Vertical Self-balancing AUV was conceived and adopted for this research work. This paper presents the development of a Vertical Self-balancing Experimental AUV and its initial experimental results. Index Terms–––Vertical Self-balancing Autonomous Underwater Vehicle, position control

I.

INTRODUCTION

The unmanned underwater vehicle (UUV) covers both remotely operated vehicles (ROVs) and autonomous underwater vehicle (AUVs). ROVs have tethered umbilical cable to enable remote operator to control the operation of the vehicle. Tether influences the dynamics of vehicle, greatly reducing maneuverability. AUVs are tethered free, unmanned, powered by onboard energy sources such as fuel cells and batteries. AUVs performing manipulation or inspection tasks need to be controlled in six degrees of freedom [1]. AUVs are also equipped with devices such as electronic compass, GPS, sonar sensor, laser ranger, pressure sensor, inclination sensor, roll sensor and controlled by onboard computers to execute complex preprogrammed missions. In the oil and gas sector, two separate classes of AUVs are identified for application in exploration and production. A survey class is for inspection of offshore structures and data acquisition, and a work class is for underwater manipulation required in installation of underwater facilities and equipments. In order for the AUV to achieve autonomy under ocean environment, the control system must have the adaptability and robustness to the nonlinearity and timevariance of AUV dynamics, unpredictable environmental uncertainties such as sea current fluctuation and reaction force from the collision with sea water creatures. Ground motion as a result of seismic activity could result in underwater slope instability and land slide. Other geohazard issue such as shallow gas hydrates poses threats to human diver. Human divers have traditionally being well suited in offshore underwater work in oil industries. Human divers in upright position exhibits larger volume to diameter ratio, and larger vertical height to horizontal length ratio

which give rise to better ability to maneuver tight corners in between pipelines, risers and other sub-sea facilities. Therefore, after the process of several design concepts generation and evaluation, the concept of revolutionary human like vertical self-balancing AUV was conceived and adopted for this research work. Cylindrical Shape could provide larger volume to diameter ratio than the spherical shape and Vertical orientation provide human like ability to maneuver tight corners in between pipelines, risers and other sub-sea facilities where the lateral space is limited. This paper discusses the development of a miniature Vertical Self-balancing Experimental AUV. This experimental AUV shall provide a low cost platform for testing the responses of this particular new type of AUV to various control algorithms.

II.

MOTIVATION

A.

Market Drivers The offshore oil industry is currently pursuing offshore oil production with well depths that previously would have considered technically infeasible or uneconomical [2]. An industry study [3], Figure 1, shows the maximum well depth versus past years. In 1949, the offshore industry was producing in about 5m water depth and it took 20 years to reach about 100m. However, in recent years the maximum well depth increases dramatically. It seems that the maximum well depth will continue to increase in the future. The maximum acceptable depth limit for a human diver is about 300m. Deeper than these depths, ocean floor mapping, inspection and repair operations of facilities must be executed by either UUVs or inhabited submersibles.

Figure 1. Offshore Oil Fields – Maximum well depth versus past years. Data and figure source: [3]

International Journal of Engineering & Technology IJET-IJENS Vol:10 No:02 B.

Future of ROVs and AUVs Due to increasing offshore oil well depth, the limits of ROVs have been reached and further advancement in UUV requires more research work on the technology of AUVs. AUVs are free from constraints of an umbilical cable and are fully autonomous underwater robots designed to carry out specific preprogrammed tasks such as ocean floor mappings, manipulation, installation and repair operations of underwater facilities.

III. LITERATURE SURVEY Literature survey is divided into two parts. Survey on Experimental AUVs and another is survey on the Previous Work on AUV control. A. Survey on Experimental AUV Experimental AUVs provide an excellent platform for development and testing of various new control methodology and algorithms to be implemented in developing advanced AUVs. In 1995, with intention to contribute to AUV development, the Autonomous Systems Laboratory (ASL) of the University Of Hawaii has designed and developed the Omni-directional Intelligent Navigator (ODIN-I). In 1995, ODIN-I was refurbished and ODIN-II was born [4]. ODIN-I and ODIN-II have made precious contribution in the development and testing of various new control methodology and algorithms. In 2003, ODIN-III was developed. It has the same external configuration and major sensors as ODIN-II, which is a closed-framed spherical shaped vehicle with eight refurbished thruster assemblies and a one DOF manipulator. ODIN-III represents Experimental robotic class AUV. A miniature cylindrical shaped AUV called REMUS (Remote Environmental Monitoring Units) is designed to conduct underwater scientific experiments and oceanographic surveys in shallow water [5]. REMUS is equipped with physical, biological and optical sensors. A standard REMUS is 19 cm in diameter and 160 cm long. REMUS represents experimental survey class AUV. Even smaller experimental class AUV was developed to demonstrate that the basic components of AUV can be packaged in 3 inch diameter tube [6]. The hull is made from standard 3 inch Schedule 40 PVC pipe. The nose cone is machined from a solid piece of PVC. The tail is made from high density foam bonded to a piece of aluminum tubing. B. Survey on the Previous on AUV Control SMC is a nonlinear feedback control scheme. Yoerger [7,8] developed SMC methodology for the trajectory control of AUV. Yoerger [9] developed an adaptive SMC for the control of experimental underwater vehicle. Song [10] proposed a Sliding Mode Fuzzy Controller (SMFC). The effectiveness of the control philosophy was tested on Ocean Explorer series AUVs developed by Florida Atlantic University. The problem with SMC is the chattering.

82

NNC is a adaptive control scheme. NNC poses parallel information processing features of human brain with highly interconnected processing elements. The main attractive features of neural networks include the self learning and distributed memory capabilities [11]. J Yuh [12] proposed a multilayered neural network for the AUV control, in which the error-back propagation method is used. The development of this scheme was motivated by J Yuh [13] which exhibit that the teaching error signal, the discrepancy between the actual output signal and teaching signal can be approximated by the output error of the control system. J S Wang [14] proposed Neuro-Fuzzy control systems. Tamaki URA [15] proposed a Self Organizing Neural Net Controller System (SONCS) for AUVs. K Ishii [16] proposed an on-line adaptation method “Imaginary Training” to improve the time consuming adaptation process of the original SONCS. Performance of NNC depends on proper tuning. PID is used for control over steady state and transient errors. PID has been widely implemented in process industries. It is also used as a benchmark against which any other control scheme is compared [17]. B Jalving [18] proposed three separate PID technique based controllers for steering, diving, and speed control. The concept was tested on Norwegian Defense Research Establishment (NDRE) AUV. The performance of the flight control system was reported to be satisfactory during extensive sea testing. J. Yuh [19] proposed Non-regressor based adaptive controller. S. Zhao [20] proposed adaptive plus disturbance observer (DOB) controller. Zhao used non-regressor based adaptive controller as an outer-loop controller and DOB as an inner-loop compensator. Zhao carried out experimental work on the proposed adaptive DOB controller using ODIN. Experimental work involves determining the tracking errors associated predetermined trajectories. Zhao [21] recommended future research works on as follows. 1.

“Survey other recently developed robust compensators such as Nonlinear Disturbance Observer” (NDOBC).

2.

“Have more tests with disturbance exerted in a more precise and controlled way”

C. Conclusion on the Literature Review Various types of Experimental AUVs are developed by various institutions or organizations to study the responsiveness or capability of their AUVs in the aspect of position control, tracking given reference position and attitude trajectories while the AUV is under the control of recently developed several advanced control algorithms. With the intension to contribute in AUV research and developments, the authors have developed a unique experimental AUV through the methodology discussed in the next section.

International Journal of Engineering & Technology IJET-IJENS Vol:10 No:02

83

IV. METHODOLOGY The flow chart of the methodology is shown in Figure 2. Methodology of the research project has started with the clarification of tasks, generation of conceptual designs of the experimental AUV, evaluation of the conceptual designs, detail design of the selected experimental AUV design concept, selection of sensors and microcontrollers, fabrication and assembly.

Development of a programmable vertical self-balancing Experimental AUV

Clarification of the tasks

Generate Conceptual Designs of AUV.

Evaluate Conceptual Designs based on selection Criteria Actual Thruster

Is one of the concepts scores better than the rest?

No

Thruster arm Counter balance

Yes Detail design of the selected Experimental AUV concept. Select microcontroller and sensors.

Fabricate and assemble

Programmable Experimental AUV & Testing Figure 2. Flow chart of the methodology A. Generation of Conceptual Design AUVs performing manipulation or inspection tasks need to be controlled in six Degree of Freedom (DOF). Therefore, any new AUV design concept must have the mechanisms to control in six DOF. Four conceptual designs were generated. All four generated conceptual designs are shown in Figure 3.

Figure 3. Four conceptual designs of AUVs B. Evaluation of Conceptual Designs Evaluation of conceptual designs was done using method described in [22]. Criteria for evaluation are Appropriate Centre of Buoyancy (CB) location (i.e. C.B as close as possible to C.G.), Ability to maneuver effectively in between Pipelines and Sub-sea Structure (i.e. larger volume to

International Journal of Engineering & Technology IJET-IJENS Vol:10 No:02 diameter ratio, larger vertical height to horizontal length ratio), Six DOF response (i.e. ability to response effectively to errors in Yaw, Roll, Surge, Sway, Pitch and Heave), Ease of fabrication (i.e. low complexity of components and low number of assembly steps) and Expandability (i.e. expandable in volume without needing to re-fabricate whole shell). Final selected concept was the concept 3 and was named a Vertical Self-balancing AUV.

cylinders. Solid aluminum alloy round bar were CNC machined to produce segments of hollow cylinder. Segments of hollow cylinders were joined to one another by male and female connection. “O” ring is sandwiched between the male and female connection to achieve water-tightness. Brace bars are used to keep the joint in place.

Holes for the thruster arm and its collar.

Figure 4. shows the sketch of the Vertical Self-balancing AUV by Computer Aided Three-dimensional Interactive Application (CATIA). Grove for “O” ring

Grove for “O” ring

Counter thrusters Actual Thruster

84

Material removal for the weight adjustments

Counter-balance Thruster Arm

Figure 5. CATIA 3D model of half of the main hull

Figure 4. Final Selected Concept – Vertical Self-balancing AUV

Figure 6. shows the AUV in the Flume tank at Offshore and Coastal Engineering Laboratory, Universiti Teknologi Petronas (UTP). The diameter of the main hull hollow segments is 0.12 m. The overall height of the vehicle is 0.83 m. The overall width of the vehicle is 0.59 m. The weight of the AUV in air is 16.3 kg.

Please Note: To balance AUV, thrusters are to rotate around the axis of the thruster arm by servo motor which is controlled by microcontroller. Pitch – Tilting of the thrusters by servo motors Yaw – Power adjustments of the two counter thrusters Heave – Power adjustments of all the four thrusters Roll – Tilting of the thrusters by servo motors

V. DESCRIPTION OF THE VEHICLE

Thruster arm

Thruster arm collar

The AUV concept was further developed with few critical specifications that have been segregated into two categories. The two categories are namely “Demand” and “Desire”. Demand: Maximum weight in air < 20 kg Maximum depth of operation < 5 m Maximum height < 0.9 m Maximum width < 0.6 m Cylindrical shaped Mode of operations: Autonomous or umbilical Desire: Able to accommodate extension in height Main Hull of the Vehicle A. Main Hull of the Vehicle Figure 5. shows the CATIA 3D model of half of the main hull. The main hall is made up of six segments of hollow

Figure 6. AUV in Flume Tank at UTP Epoxy is used to form a permanent joint between the main hull and thruster arm collar. “O” ring is sandwiched between the thruster arm collar and the thruster arm. The weight of the AUV is adjusted to be slightly positive buoyant so that in the event of the power failure it has the ability surface on its own.

International Journal of Engineering & Technology IJET-IJENS Vol:10 No:02 B. Thruster System The vehicle has four vertical thrusters and two horizontal thrusters. All the thrusters can turn in either clockwise or anticlockwise. Figure 7. shows a photograph of a fully assembled thruster. Servoflex Coupling is used to couple front end of the thruster motor shaft and propeller shaft. Servoflex Coupling allows for minor misalignment of the shaft. Shaft encoder is coupled to the rear end of the thruster motor shaft. Pulses from the shaft encoder are fed in to the slave microcontroller based control system to keep the motor rpm as commanded by the master microcontroller.

Thruster arm Thruster cap flange

Thruster housing

Thruster base flange

Propeller shaft

Figure 7. The photograph of a fully assembled thruster The diameter of thruster housing is 0.088 m. “O” ring is sandwiched between the thruster housing and thruster base flange. “U” shaped standard oil seal is used to seal the water between the propeller shaft and the thruster base flange. Figure 8. shows a mechanism to ensure that thruster assembly does not revolve around the thruster arm when the thruster motor spins. Epoxy is used to form a permanent joint between the thruster jam ring and the thruster cap flange. There are two M3 taps at the thruster jam ring. M3 cap screw passes through the thruster jam ring and through the 3 mm hole of the thruster arm. “O” ring is sandwiched between the thruster housing and thruster cap flange. There is also an “O” ring sandwiched between the thruster arm and thruster cap flange. Thruster cap flange Thruster jam ring “O” Ring

C. Control System of the Vehicle Accelerometer is used to sense the roll and pitch angle. Gyro is used to sense the yaw rate angle. For this experimental AUV, the sonar sensor is used to measure the horizontal translation displacement, which is sway and surge. The pressure sensor is used to sense the vertical displacement, which is depth. The main microcontroller used is the Rabbit RCM2000. It has twenty six configurable inputs and outputs, eight fixed inputs and six fixed outputs. Control programs for the RCM2000 are written on the personal computer. The software used to write the control programs is called “Dynamic C”. Dynamic C is an integrated C compiler, editor, loader and debugger. Analog to Digital converter was used to interface the RCM2000 with analogue sensor. D. Power requirement of the Vehicle Two types of power loads are typically identified by the AUV designers. One is propulsion load (Thrusters and control surface actuators loads) and the other is hotel load (microcontrollers). Each thruster motor (12V) at maximum efficiency consumes the current of 5.16Ah. At maximum efficiency, the torque is 14.7mNm and the output power is 33.02W. Each servo to rotate the thruster consumes 1.2Ah. Hotel load is 5V and sufficient with less than 1Ah. Therefore, at maximum load, total current requirement for six thruster motors and 2 servo motors and the hotel load is about 33Ah. VI. EXPERMENTATION Testing of new AUV concept involves testing position control in six Degree of Freedom (DOF). Benchmark PID controller was used for initial testing of this experimental AUV. This controller was tested for four DOF position control. Four DOF position control tested were yaw, roll, sway and heave. Only the 4 DOF position control experimentation is required to test because roll and sway are the same as pitch and surge for this particular AUV. The input-output relationship of a PID controller is expressed as follows: 1 ⎡ 1 de(t ) ⎤ ⎢ ⎥ --- (1) e(t )dt + Td u(t) = Kc e(t ) + Ti dt ⎥ ⎢ 0 ⎣ ⎦

∫

Kc = controller gain Ti = integral time Kd = derivative time Taking the Laplace transform of equation (1), the transfer function of PID can be written as:

M3 Screw

M3 Screw Figure 8. Thruster cap flange assembly

85

K U (s ) = K c + c + K c Td s --- (2) E (s ) Ti s

International Journal of Engineering & Technology IJET-IJENS Vol:10 No:02 By obtaining the z-transform of the equation (2), the discrete form of the PID controller can be derived:

(

D( z ) = K c +

)

K cT

(

Ti 1 − z

−1

)

(

C. Results and Discussion of the Experiments

)

Tracking of Yaw (Result of PID Controller without Disturbance)

⎤ ⎥ ------ (3) ⎥⎦

)

K T 1 − z −1 + c d ------ (4) T

0.3 0.2 Yaw (radian)

(

⎡ 1 − z −1 U (z ) T T = K c ⎢1 + + d −1 E(z ) T ⎢⎣ Ti 1 − z

86

0.1 0 -0.1

0

20

40

60

80

100

120

-0.2 -0.3 Time (Second)

The transfer function in equation (4) was implemented as parallel structure by summing up the proportional, integral and derivative terms.

Tracking of Yaw (Result of PID Controller with Disturbance)

Yaw Angle (Radian)

0.3

A. Experimental Settings In the experiment, accelerometer was used to measure the roll angle and gyro is used to measure the yaw angle. This angular measurement sensor system is most commonly known as Inertia Navigation System (INS). The sonar sensor was used to measure the horizontal translation displacement, which is sway. The pressure sensor was used to measure the vertical displacement, which is depth. The signal from the INS, sonar and pressure sensors were also used to feed into the controller. The sampling period from the INS and sensors was 0.5 second.

0.2 0.1 0 -0.1

0

20

40

60

80

100

120

-0.2 -0.3 Time (Second)

Figure. 9. Tracking of Yaw using PID Tracking of Roll (Result of PID Controller without External Disturbance)

The angular, vertical and translational positions of the AUV were recorded in real time using a high resolution (24 Bit) data logger unit and software. The angular, vertical and translational positions of the AUV were recorded in voltage. In the case of angular measurements, calibration Slope of voltage to radian was developed. Then, the voltage reading was interpreted to radian. Then the graph of radian against time was plotted.

Roll Angle (Radian)

0.15 0.1 0.05 0 -0.05 0

20

40

60

80

100

120

-0.1 -0.15

Time (Second)

Tracking of Roll (Result of PID Controller with External Disturbance)

Ro ll An g le (Rad ian )

0.15 0.1 0.05 0 -0.05

0

20

40

60

80

100

120

-0.1 -0.15

Time (Second)

Figure. 10. Tracking of Roll Using PID Tracking of Heave (Result of PID Controller without Disturbance) Heave (Metre)

B. Brief Description of the Experimental Procedure Initial tests on this new type of AUV were done by conducting position control experiments using benchmark controller (PID). Yaw control experiment was started after the AUV was manually turned to a fixed amount of angle from the nominal (zero) position. The requirement of the experiment is zero steady state error. For this yaw control experiment, other five DOF has been constrained by the test rig. Two sets of experiments were done. One is without the disturbance and another is with disturbance. Disturbance is achieved by manually turning the AUV to a fixed amount of angle from the nominal (Zero) position. The same procedure was repeated for roll control experiments except that other DOFs were not constrained and roll angle was not turned manually at the start of roll experiment. The delay time, rise time, peak time, maximum overshoot and settling time of the controller was then obtained.

0.9 0.85 0.8 0.75 0

20

40

60

80

100

Time (Second)

Figure. 11. Tracking of Heave using PID controller

120

International Journal of Engineering & Technology IJET-IJENS Vol:10 No:02

Acknowledgements

Tracking of Sway (Result of PID Controller without Disturbance) S w ay (M etre)

87

The authors are thankful to Universiti Teknologi PETRONAS for granting them the necessary fund and facilities to execute this work.

0.8 0.6 0.4 0.2 0 0

20

40

60

80

100

120

References

Time (Second)

Figure. 12. Tracking of Sway using PID controller Figure. 9 show the tracking of yaw. For the step response, the average delay time is 2.3s, the rise time is 3.5s, the peak time is 5.8s and settling time is 14s. When the external disturbance is introduced, PID detects the position error and corrects it through feedback, and hence keeps up its tracking performance. Disturbance was introduced at about 63 seconds. However, position tracking performance decay after the external disturbance is introduced. This may be due to the fixed controller gains as opposed to controller gains are tuned automatically as in the case of adaptive controllers. Figure. 10 show the tracking of roll. Disturbance was introduced at about 76 seconds. The result shows similar trend as in the case of yaw. Figure. 11 show the tracking of heave. No disturbance was introduced in this heave test. There were fluctuations in tracking performance. Figure. 12 show the tracking of sway. No disturbance was introduced in this heave test. There was the decay in tracking performance even without the external disturbance. The results may be improved by testing the AUV with more robust controllers such as position error observer based controller or adaptive DOB controller that has been proposed by Zhao [21].

VI. CONCLUSION Experimental type Vertical Self-balancing AUV has been designed, fabricated and assembled. This experimental AUV is to provide a low cost platform for testing the responses of this particular type of AUV to various control algorithms. Initial tests on this new type of AUV were done by conducting position control experiments using benchmark controller (PID). Since the AUV has been designed and the components has been sourced by the end users themselves, if there is any breakdown of components the users will be fully aware of where to source for the replacement parts and how to replace them. Therefore, the down time will be reduced to minimum. The vehicle is also able to accommodate extension of the main hull of the vehicle should there be the need to increase height of the vehicle. The hardware design of the AUV can also accommodate to any modifications required to test it in different orientation. For example, the same main hull of vehicle can also be used to construct a conventional horizontally orientated AUV with tail fins and tail propeller. Nose cone and tail section with the propeller and tail fin can be fixed to the same main hull with ease.

[1] Gianluca Antonelli, Stefano Chiaverini, Nilajan Sarkar, and Michael West, “Adaptive Control of an Autonomous Underwater Vehicle: Experimental Results on ODIN” IEEE Transaction on Control Systems Technology, Vol. 9, No. 5, pp. 756 – 765, September 2001. [2] Loius L. Whitcomb, “Underwater Robotics: Out of the Research Laboratory and Into the Field” International Conference on Robotics and Automation, IEEE 2000. [3] D. Harbinson and J. Westwood, “Deepwater Oil & Gas – An Overview Of The World Market,” Deep Ocean Technology Conference, New Orleans, 1998. [4] H.T. Choi, A. Hanai, S.K. Choi, and J. Yuh, “Development of an Underwater Robot, ODIN-III,” Proceedings of the 2003 IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 836 – 841, Las Vegas, Nevada, October 2003. [5] Mike Purcell, Chris von Alt, Ben Allen, Tom Austin, Ned Forrester, Rob Goldsborough and Roger Stokey, “Nee Capablities of the REMUS Autonomous Under Vehicle,” IEEE 2000. [6] Aditya S. Gadre, Jared J. Mach, Daniel J. Stilwell, Carl E. Eick, “Design of a Prototype Miniature Autonomous Underwater Vehicle,” Proceedings of the 2003 IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 842 – 846, Las Vegas, Nevada, October 2003. [7] D. Yoerger, J. Newman, “Demonstration of closed-loop Trajectory Control of an Underwater Vehicle,” 1985 Proceedings of OCEANS, vol 17, pp. 1028 – 1033, Nov. 1985. [8] D. Yoerger, J. Slotine, “Robust Trajectory Control of Underwater Vehicles,” IEEE Journal of Oceanic Engineering, Vol. OE-10, No.4, pp. 462 – 470, October 1985. [9] D. Yoerger, J. Slotine, “Adaptive Sliding Control of and Experimental Underwater Vehicles,” Proceedings of the 1991 IEEE International Conference on Robotics and Automation, pp. 2746 – 2751, Sacramento, California, April 1991. [10] F. Song and S. Smith, “Design of Sliding Mode Fuzzy Controllers for Autonomous Underwater Vehicle without System Model”, OCEANS’2000 IEEE/MTS, pp. 835-840, 2000.

International Journal of Engineering & Technology IJET-IJENS Vol:10 No:02 [11] Arthur G.O. Mutambara, “Design And Analysis of Control Systems” CRC Press 1999. [12] J Yuh, “A Neural Net Controller for Underwater Robotic Vehicles,” IEEE Journal of Ocean Engineering, Vol. 15, No. 3, pp. 161 – 166, July 1990. [13] J. Yuh, R. Lakshmi, S. J. Lee, and J. Oh, “An Adaptive Neural-Net Controller for Robotic Manipulators,” in Robotics and Manufacturing, M. Jamshidi and M. Saif, Eds. New York: ASME, 1990. [14] Jeen-Shing Wang and C. S. George Lee, “Efficient Neuro-Fuzzy Control Systems for Autonomous Underwater Vehicle Control,” Proceedings of the 2001 IEEE International Conference on Robotics and Automation, pp. 2986 – 2991, Seoul, Korea, 2001. [15] Tamaki URA, Teruo FUJII, Yoshiaki NOSE and Yoji KURODA, “Self-Organizing Control System for Underwater Vehicles,” IEEE 1990. [16] Kazuo Ishii, Teruo Fujii, and Tamaki Ura, “An On-line Adaptation Method in a Neural Network Based Control System for AUV’s,” IEEE Journal of Ocean Engineering, Vol. 20, No. 3, pp. 221 – 228, July 1995. [17] Ahmad M. Ibrahim, “Fuzzy Logic for Embedded Systems Applications” Elsevier Science 2004. [18] BjYrn Jalving, “The ADRE-AUV Flight Control System,” IEEE Journal of Ocean Engineering, Vol. 19, No. 4, pp. 497 – 501, October 1994. [19] J. Yuh, Michael E. West, P.M.Lee, “An Autonomous Underwater Vehicle Control with a Non-regressor Based Algorithm +,” Proceedings of the 2001 IEEE International Conference on Robotics and Automation, pp. 2363 – 2368, Seoul, Korea, May 2001. [20] S. Zhao, J. Yuh, and S. K. Choi, “Adaptive DOB Control for AUVs,” Proceedings of the 2004 IEEE International Conference on Robotics and Automation, pp. 4899 – 4904, New Orleans, LA, April 2004. [21] Side Zhao, “Advanced Control of Autonomous Underwater Vehicles”, PhD Thesis in Mechanical Engineering, University of Hawaii, August 2004. [22] David G. Ullman, “The Mechanical Design Process”, Mc Graw Hill, Third Edition, International Edition 2004.

88