A Quasi-Stabilized Underwater Remotely Operated Vehicle Clement Ong, Francisco Guzman II, Marc Legaspi, Cathleen Faith Tan, Jolly Ann Uy College of Computer Studies, De La Salle University-Manila 2401 Taft Avenue, Malate, Manila Tel: +63 2 5240402, +63 2 5360278

[email protected] ABSTRACT A prototype small-scale propelled submersible craft capable of quasi-stationary attitude control has been developed. The submarine is cylindrical in shape and uses two thrust-producing propellers driven by DC motors. Variable volume ballast tanks are servo-controlled against an inertial measurement unit (IMU) to provide automatic attitude stabilization based on remotely-input user demand. In this paper the overall design of the prototype is presented, with emphasis on the attitude and ballast control system. Performance results are shown and discussed, and conclusions and recommendations are given.

Keywords Underwater ROV, Attitude Control, Buoyancy Control, Inertia Measurement

1 INTRODUCTION Underwater Remotely operated vehicles (ROVs) are mobile tools that have been used for the last 30 years to get access to locations that are inaccessible to divers. A typical ROV might consist of a camera and a two way communications mechanism (a tether or umbilical cord) that allows the remote operator to control the vehicle. ROV applications also include a range of military, commercial and scientific needs, such as unmanned underwater vehicles (UAVs) and remotely controlled bomb disabling vehicles [21]. ROVs typically require constant human supervision due to the unpredictability of the aquaeous environment, such as changing currents and pressure. For certain operations such as surveying and salvage / retreival work, a steady craft is a necessity – putting heavy demands on the operator. To reduce operator fatigue and to allow for longer mission times, submersible device should be able to correct itself in terms of position and direction.

2 PROTOTYPE DESIGN 2.1 Overview SeaROV is a small-scale prototype craft whose cylindrical hull is fabricated using 64mm acrylic plastic (Figure 1). The hull is 60cm long, and 16.5cm in diameter. Four independently-variable volume ballast tanks and two DC-motor driven propellers are combined to provide thrust and attitude control.

Figure 1. SeaROV acrylic body with fixed weights for ballast. Communications and power are provided to the craft via a tether. User input and data to/from the ROV are facilitated through a PC with custom-written software, which also provides data logging capabilities. The PC keyboard and a joystick provide the user interface to piloting the vehicle. The system block diagram is shown in Figure 2. The user inputs commands on the User Control and Communication Module. The information is sent to the Intercommunication Module, which in turn relays instructions to the Propulsion Module and the Ballast Module. As feedback, the Ballast Module outputs the current water content of the ballast tanks to the Intercommunication Module. The Inertia Module provides the Intercommunication Module with the current information of the attitude and inertia state of the SEA-ROV while it is in operation. This information is sent to the User Control and Communication Module. The User Control and Communication Module then displays the current attitude and depth of the vehicle. Based on the commands of the user, the information received by the Intercommunication Module would force the Ballast Module and Propulsion Module to operate appropriately.

2.2 Attitude Control Software Given the commands from the User Control and Communication Module, the SEA-ROV follows preset motor and ballast activity instructions during operation. The software checks if the state of the ballast tank is to be changed. If the ballast tank state needs to be modified then the software issues a ballast control selector signal for specifying the appropriate ballast tanks to manipulate and the corresponding intake/discharge rates to the Ballast Module. If the commands require a change in the states of the propellers then the software sends the consequent rotation rate and direction to the Propulsion Module.

controls the forward pitch of the vehicle, while combination of B and D controls the backward pitch. The combination of ballast units A and B controls the rightward roll, while the combination of C and D controls the leftward roll. References to ballast design can be found in [11], [12], [14], and [20].

User

Interaction Software

Acceleration Measurement Submodule

Acceleration Inertia State Mode Input Data

Depth

Depth and Inertia Computation Software

X Axis State Y Axis State

Command

Inertia State

PC Communication

Attitude Measurement Submodule

Z Axis State Inertia State

Acceleration

Command

Initial Y Axis State

User Control and Communication Module

Relay

Initial Y Axis State Initial X Axis State

Intercommunication Module Direction Propulsion Module

Ballast Level Intake/Discharge Initial Rate Control State

Throttle

Ballast Module

Motor and Ballast Control Software

Throttle

Inertia Module

Calibration Software

Direction Intake/Discharge Rate

Propulsion Module

Level

Ballast Module

Figure 4. Ballast tanks A, C versus B, D control pitch, while A, B versus C, D control roll.

Ballast Control

Figure 2. System Block Diagram of SeaROV

2.3 Ballast Module The Ballast Module (Figure 3) consists of four independently controlled ballast tanks that allow the SEA-ROV to rotate along the x and z axes and at the same time controlls the craft's buoyancy. The module manipulates the contents of the ballast tanks indicated by the ballast control signal and floods/empties these tanks based on a given intake/discharge rate to induce the desired tilt. As feedback, the ballast module returns the amount of water contained in the ballast tank back to the Intercommunication Module. Intercommunication Module

2.4 Inertia Module The Inertia Module oversees the inertia data acquisition essential to SEA-ROV’s operation. This module supplies the Intercommunication Module with the current acceleration and states of the x, y and z axes. The Attitude Measurement Submodule measures the x, y and z axis states while the Acceleration Measurement Submodule measures the acceleration along these axes. These accelerations, when integrated, would result into velocities which, in turn can be used along with time to compute for the distance on each axis. Z Axis State

Intercommunication Module

Y Axis State

Attitude Measurement Submodule

X Axis State

Ballast Control

Ballast Module

Intake/Discharge Rate

Level

Acceleration

Acceleration Measurement Submodule

Figure 3. Ballast Module Block Diagram Determining the level of water accurately inside the ballast tanks is essential since neutral buoyancy is critical, I.e. a small difference in ballast weight changes the craft from neutral to positive or negative buoyancy. The ballest tank volumes are controlled by DC motors driving screw shafts, which actuate a piston in each ballast tank. Rotary encoders are used to keep track of icnremental rotations to provide a means of indirectly measuring each tank's volume. Figure 4 is the physical arrangement of the Ballast Tanks inside the SEA-ROV. Letters A, B, C and D represent the ballast tanks of the SEA-ROV. The combination of ballast units A and C

Figure 5. Inertia Module Block Diagram

2.4.1 Acceleration Measurement Submodule The Acceleration Measurement Submodule applied in the system measures the static acceleration of gravity, as well as dynamic acceleration resulting from motion, shock, or vibration. In order to measure the acceleration from the x, y, z axis, a tri-axial accelerometer used, with a ± 3 g sense range. A quad rail to rail operational amplifier buffers the output from the accelerometer chip, greatly reducing output impedance.

2.5 Control System The prototype currently uses the simplest closed-loop control system – relay-operated on/off is used for thrusters drive limiting the control algorithm to simple limit-cycle approach. The ballast control uses a slightly more sophisticated feedforward controller (implemented by table-lookup) to control pitch and roll, as well as craft buoyancy. A much more advance control system that is promising is in [4], which can be considered for future work. Figure 6. Acceleration Measurement Submodule Schematic Diagram The 3D analog accelerometer used is the DE-ACCM3D from Dimension Engineering. The DE-ACCM3D uses a 3D accelerometer from Analog Devices, which is the ADXL330. The sensitivity of the accelerometer is 333mV/g. The RMS noise is typically 7.3mg, and output bandwidth is 500Hz. The full specification of this accelerometer is in [13].

2.4.2 Attitude Measurement Submodule SEA-ROV uses piezoelectric gyroscopes in order to determine the attitude of the vehicle. By inputting the sensors axis bits, the attitude of the vehicle can be measured. The capacitors per gyroscope is used as decoupling capacitors in order to limit the drift of the gyroscope due to the input voltage.

3 PERFORMANCE All tests were carried out in a swimming pool where the maximum depth is limited to approximately 1.6m.

3.1 Buoyancy Tests Figure 8 is the graph of the depth versus the amount of water in the ballast tanks. The relationship between water intake values and depth are proportional and critical. At around 125cc, the craft will start to submerge. At a depth level between 0 (I.e. floating on surface) and 40cm, the resultant depth versus ballast volume is not as linear. 160 140

Depth (cm))

120 100 80 60 40 20 0 0

20

40

60

80

100

120

140

160

180

Am ount of Water (cc)

Figure 8. Correlation Between Water Intake and Depth of the SEA-ROV

Figure 7. Gyroscopes Schematic Diagram The circuits under each Gyroscope IC are active pass band filters that amplify and filter the outputs of each gyroscope. The purpose of these circuits is to increase the sensitivity of the system and minimize errors due to external factors. The VDD in the circuit is the voltage inputted into each gyroscope. The voltage is divided by half by the resistor network and is used in order to closely center the signal close to 0, since the output of the gyroscope is centered around half of the supply voltage.

During this point in the test, the vehicle was oscillating up and down, due to the movement of the water. At 161cc of ballast, the vehicle had reached the bottom of the testing area. Further testing in the range of 125cc to 160cc of ballast was undertaken to determine the transition points between positive, neutral and negative buoyancy.

1.5

40 30

1

10 Angle

Buoyancy y

20 0.5 0 0

20

40

60

80

100

120

140

160

180

0 -10 0

-0.5

10

20

30

40

50

60

-20 -30

-1

-40

-1.5

-50

Am ount of Wate r (cc)

Time

Figure 9. Ballast Volume and Critical Buoyancy In Figure 9, it can be seen that as water increases inside the ballast tanks the buoyancy would change from positive to negatively buoyant with only a small window in between for neutral buoyancy. Water from 125 cc to 159 cc maintains the SEA-ROV in a neutrally buoyant state. This can be modified as necessary by the addition (or removal) of fixed weights within the craft as might be needed if additional equipment or a different buoyancy is experienced (i.e sea water).

Figure 11. Graph of the Test Results on the Closed-looped Attitude Control for Pitch (30°)

70 60 50 40

Angle

30

3.2 Closed-looped Attitude Control System Performance

20 10 0 0

3.2.1 Attitude Control for Pitch

20

40

60

80

100

120

140

160

180

-10 -20

60 50 40 30 20 10 0 -10 0 -20 -30 -40 -50

-30 Tim e

Figure 12. Graph of the Test Results on the Closed-looped Attitude Control for Pitch (45°)

Angle

Angle

Pitch requirements of 15º, 30º, 45º, 60º, and 75º as well as their negative counterparts are commanded to the craft, and the results were recorded. Figures 10 to 14. provide a visual as to the craft's responses.

50

100

150

80 70 60 50 40 30 20 10 0 -10 0 -20 -30

10

20

30

40

50

60

Time

Time

Figure 10. Graph of the Test Results on the Closed-looped Attitude Control for Pitch (15°)

Figure 13. Graph of the Test Results on the Closed-looped Attitude Control for Pitch (60°)

25

80

20

60

15

Angle

40

10

20 5

0 0

10

20

30

40

50

60

-20

0 0

10

20

30

40

50

60

-5

-40

-10

Time -15

Figure 14. Graph of the Test Results on the Closed-looped Attitude Control for Pitch (75°) The graphs clearly illustrate the oscillation in the sensor readings as the closed-loop control system tries to stabilize the system to satisfy the given pitch. Majority of the tests have a transition time of less than one minute to reach the specified angle. It can be observed that the SEA-ROV encounters different ranges of overshoot. The time it takes for the SEA-ROV to stabilize itself after drawing in or releasing water from the ballast tanks actually depends on the total amount of water taken in; the sensitivity of the vehicle increases as it nears a neutrally buoyant state. Similar results were observed for negative pitch angles, and it was determined that the maximum pitch angle was limited to a 80º to -80º. Without the help of momentum and/or oscillation induced by the vehicle’s movements the SEA-ROV is not able to reach a 90º or a -90º pitch. The SEA-ROV is capable of pitching to the target value within a time range of 50 seconds to 165 seconds.

3.2.2 Attitude Control for Roll A specified roll betweem 5º and 30º as well as their negative counterparts is commanded to the craft, and the corresponding response recorded and shown in Figures 15 to 17. 8

Time (s)

Figure 16. Graph of the Test Results on the Closed-looped Attitude Control for Roll (15°) 40 35 30 25 20 15 10 5 0 0

10

20

30

40

50

60

70

-5 -10 Time

Figure 17. Graph of the Test Results on the Closed-looped Attitude Control for Roll (30°) Due to the ballast tank arrangement being nearer to the fulcrum center of the hull and overall weight of the SEA-ROV, the vehicle can only reach a maximum of 30º and -30º roll. Since the activated ballast tank pairs (pairs A, C and B, D) are situated nearer to the center of buoyancy the overshoot values are less compared to that of the pitch angle. The craft responded and reached the target angles in 25 to 60 seconds.

6

3.2.3 Attitude Control for Yaw

4

Angle

2 0 0

5

10

15

20

25

30

-2 -4 -6 -8 Tim e (s)

Figure 15. Graph of the Test Results on the Closed-looped Attitude Control for Rol1(5°)

A specified yaw is inputted into the software. The SEA-ROV then manipulates its orientation depending on the input angle through its bidirectionally-driven propellers. A target yaw of -45 and +90 degrees is given. The results are graphically presented in Figures 18 and 19.

compass (I.e. fluxgate), which is essentially driftless, can be used to stabilize the yaw control of the vehicle.

5 0 -5 0

5

10

15

20

25

For the ballast design, a more voluminous ballast tank is suggested. This results into greater variable weight, therefore enabling the craft to perform higher angles of pitch and roll during attitude control operations. To further increase the performance of the ballast system, ballast tanks that can be filled up and emptied in shorter time durations must be implemented. A separate, smaller ballast tank positioned in the center of the hull can also be added for better, more accurate depth trimming as the vehicle’s buoyancy becomes significantly more sensitive as it nears neutral buoyancy.

-10 Angle

-15 -20 -25 -30 -35 -40 -45 -50 Time (s)

Angle

Figure 18. Graph of the Test Results on the Closed-looped Attitude Control for Yaw (-45°) 100 90 80 70 60 50 40 30 20 10 0 -10 0

As small-scale submarines continue to be researched, what has been learned in mobile robotics for autonomous navigation – simulation, localization and mapping [9], [1], [2] – is now being applied. We can only look forward to advancements in underwater robotics in the future. .

5 REFERENCES

5

10

15

20

25

Time (s)

Figure 19. Graph of the Test Results on the Closed-looped Attitude Control for Yaw (90°) Figure 18 shows the craft moving toward the required angle, stopping at -40 degrees, and slowly drifting to -45 degrees. Figure 19 shows the craft going toward the +90 degree yaw, however the movement is not asymptotic to the required position as seen by the steep angle of the graph. This error is due to the gyroscope being unable to detect the slow angular velocities induced as the vehicle nears the specified angle, showing the limit of performance of the IMU module.

4 CONCLUSIONS, RECOMMENDATIONS A prototype submersible ROV has been developed with quasistationary attitude control. The resulting prototype performance shows useable pitch and roll attitude control of the vehicle, while the yaw control is limited by the inertial measurement system. The ballast system allows depth control of the vehicle from positive to neutral to negative buoyancy. Although not shown here, the craft is agile and fairly responsive, with the operator control software allowing the user to run the vehicle in “free” as well as feedback controlled modes. Improved control-system performance is expected by utilizing high-frequency pulse width modulation (PWM) in the ballast and propulsion system, as against the “on-off” relay control currently employed. The accelerometer, DE-ACCM3D, with its 3g sensitivity was found to be too insensitive to detect the changes in depth of the vehicle. An accelerometer with a sensitivity of better than 1g is recommended for this type of operation. A digital

[1] Burgard, W. et al. Estimating the Absolute Position of a Mobile Robot Using Position Probability Grids. [online]. Available: www.dccia.ua.es/dccia/inf/ asignaturas/RA/sesion5/burgard.aaai96.pdf. (January 30, 2006). [2] Drolet, L. et al. An Adaptable Navigation System for an Underwater ROV. [online]. Available: http://www.gel.usherbrooke.ca/laborius/papers/ISR00rov.pdf . (February 21, 2006). [3] Heiszwolf, Johan. (2001). Submarine Dive Technology. [online]. Available: http://www.heiszwolf.com/subs/tech/tech01.html. (May 25, 2006). [4] Innocenti M. et al. Robust Control Underwater Vehicles: Sliding Mode vs. LMI. [online]. Available: http://www2.cemr.wvu.edu/~gcampa/pdf/acc99.pdf. (January 29, 2006). [5] Laine, J. et al. SubjuGator: A Highly Maneuverable, Autonomous Underwater Vehicle. [online]. Available: http:// citeseer.ist.psu.edu/cache/papers/cs/10655/http:zSzzSzwww. me.ufl.eduzSz~dknzSzpaperszSzicra2000_sub.pdf/laine00su bjugator.pdf. (January 28, 2006). [6] MacPherson, D. Small Propeller Cup: A Proposed Geometry Standard and a New Performance Model. www.abcprops.com/proppdf/PropellerCup.pdf. (July 8, 2006). [7] Martin, G. (February 2005). Design of a Semiautonomous Biomimetic Underwater Vehicle for Environmental Monitoring. [online]. Available: www.netbell.ee/listak_cira2005.pdf. (February 15, 2006).

[8] McAllster, D. et al. Pumps. [online]. Available: http://www.rpi.edu/dept/chem-eng/Biotech-Environ/PUMPS/ pump.html. (July 8, 2006). [9] Roeckel, M. et al. Simulation Environments for the Design and Test of an Intelligent Controller for Autonomous Underwater Vehicles. [online]. Available: www.informscs.org/wsc99papers/158.PDF. (January 30, 2006). [10] Streeter, V. et al. Fluid mechanics. Singapore: McGraw-Hill Inc, 1998. [11] Vasilescu, I. et al. Autonomous Modular Optical Underwater Robot (AMOUR) Design, Prototype and Feasibility Study. [online]. Available: groups.csail.mit.edu/drl/ publications/papers/ICRA05b.pdf (January 30, 2006). [12] Wolf, M. et al. Dynamic Buoyancy Control of an ROV using a Variable Ballast Tank. [online]. Available: http://web.mit.edu/miwolf/www/rovpaper.pdf. (January 30, 2006). [13] Dimension Engineering. DE-ACCM3D Datasheet. [online]. Available: http://www.dimensionengineering.com/datasheets/DEACCM3D.pdf. (June 3, 2006). [14] Flubber2000. In Implementation Ballast System. [online]. Available: http://nocwebb.ntu.edu.sg/toystory/2000/flubber %202000/ihpt/tech_implement_ballast_mea.htm. (July 8, 2006).

[15] FREESUB. General Information on AUVs. [online]. Available: http://www.cybernetix.fr/freesub/general.htm. (January 28, 2006). [16] Innovative Technology. Remotely Operated Vehicle (ROV) System for Horizontal Tanks. [online]. Available: http://apps.em.doe.gov/OST/pubs/itsrs/itsr2086.pdf. (February 21, 2006). [17] ISE Group Companies. AUV DesignInfo Page. [online] Available: http://www.ise.bc.ca/WADEwhatisanAUV.html. (February 16, 2006). [18] NOAA’s Aquarius. Designing an Autonomous Underwater Vehicle. [online]. Available: http://www.uncw.edu/aquarius/ education/lessons/Aq%20AUV.pdf. (February 15, 2006). [19] Society for Underwater Technology. Underwater Robotics. [online]. Available: http://www.sut.org.uk/urghome/currapps2.html. (January 28,2006). [20] The Association of Model Submarines. Ballast Tank Design. [online]. Available: http://www.modelsubmarines.org/ballast %20tank.htm. (January 29, 2006). [21] Rov.net – ROV’s for Beginners [online]. Available http://www.rov.net/pages/rovfd1.html (Feb 17, 2006)