€¦ · web viewin addition to wireless communications, the asv also provides controllable...

Naval Engineers Journal, September 2017, Volume 129, Number 3, pp. 75-88

Andrew C. Gray, Eric M. SchwartzMachine Intelligence Laboratory

University of Florida

The NaviGator Autonomous Maritime System

ABSTRACTTransferring data between the ocean surface and an operating Autonomous Underwater Vehicle (AUV) is a difficult and slow process. Current methods of un-tethered submersible communication have either a very limited data transfer rate or a prohibitively short range. In this paper, students of the University of Florida propose a new data transfer method that will supply the high data rates of a tethered system yet still maintain the advantages of a fully autonomous submersible. This paper reports the process of designing, testing, and proving this new approach. This method will improve communication between AUVs and the operator and enhance AUV operational efficiency and safety.

This approach uses an AUV paired with an Autonomous Surface Vehicle (ASV) that carries a Remotely Operated Vehicle (ROV). During operation, the ROV is launched from the ASV. The ROV dives to the approximate operating depth of the AUV. The AUV is acoustically summoned on an as-needed basis or based on a predefined schedule. When the transducer from the ROV begins transmitting, the AUV acquires the acoustic signal. Once in visual range, visual indicators on the ROV guide the AUV toward the docking mechanism on the ROV. When docked, the AUV transfers data through the ROV and to the ASV through the tether. Data on the ASV is stored onboard or transmitted to the user through wireless or satellite communications. Once the ASV data transfer is complete, the AUV is released from the ROV to continue its normal operation.

INTRODUCTIONAutonomous Underwater Vehicles (AUVs) are part of a growing industry and have proven to be useful in maritime operations. AUVs have been used to find underwater mines, inspect oil

derricks, map parts of the ocean floor, and even search for shipwrecks. Recently, AUVs have been used to search for the missing Malaysian Airlines Flight 370 (MH370).

During the search, an underwater signal was detected that may have originated from MH370’s flight recorder. Bluefin-21 drones, shown in Figure 1, were launched from a surface vessel to search for the submerged aircraft. Every day the submersible spent 16 hours at the operating depth, four hours diving and surfacing, and another four hours transferring the data to the operator once recovered [1].

Figure 1. Recovery of a Bluefin 21 AUV [2].

Beacons used to locate the flight data recorders of downed aircraft are battery operated and must be located quickly after an accident. As MH370’s flight recorder battery waned, the need to receive data rapidly from the AUVs became critical. In addition to the difficulties of scanning the ocean floor for an aircraft or its wreckage, the overall cost of the search increased quickly after each day of searching. Within the first month, over 19 surface vessels [3] from more than 20 different countries contributed to the search. Time Magazine estimated the first three months of the search

1


cost more than $70 million [4]. The loss of the aircraft and the ensuing search proved that AUV operations played a key role. However, with the costs of operating ships and AUVs during the search, it is evident that efforts to make the operations more efficient and affordable are warranted.

EXISTING TOOLS AND METHODS

Communicating with submersibles is difficult and complicates underwater operations. One of the tools AUV operators can use to monitor the vehicles is acoustic modems. Acoustic modems are omni-directional, support extensive range, and are not susceptible to environmental conditions such as water clarity. However, most production-grade acoustic modems can only achieve data rates up to 50 Kbps [5]. Due to the attenuation experienced by sound waves traveling through the water, modems with higher data rates have transmission ranges that are too short to be useful for AUVs.

During AUV operations, human operators must be on location and actively involved. In addition to supervising the launch and recovery processes, operators must also observe the submarine’s underwater status. Operators on the surface typically monitor the data via acoustic modems. For most AUV operations, the imaging SONAR is the most useful sensor. The active SONAR provides visual data of the ocean floor such as the image shown in Figure 2. However, the active SONAR data rate can be as high as 10 Mbps [6]. At the acoustic modem’s maximum 50 Kbps data rate, more than three minutes would elapse before receiving one second of data from the imaging SONAR. Acoustic modems are too slow to keep up with the demands of real-time SONAR data transfer.

Figure 2. Imaging SONAR data showing a shipwreck on the bottom of the ocean [7].

Cameras are another useful underwater sensor. However, cameras can require an even higher data transfer rate than SONAR, depending on the camera’s specifications. To support the immense amount of data recorded by the submarine, most AUVs require the submersible to be recovered first. Operators must then download the recorded data directly from the submarine. Depending on the time spent underwater, the data could be more than 16 hours old. When timing is critical, as in the search for MH370, 16 hours of data delay is unacceptable.

This paper proposes a new system to improve AUV communication with surface users. Ideally, the new system will also reduce the need for frequent human interventions.

REMOTELY OPERATED VEHICLES AND AUTONOMOUS SURFACE VEHICLES IN AUV OPERATIONSOne way to solve the communication issue between the surface and the submersible is to use a tether. ROVs, like the one shown in Figure 3, are submarines that have attached tethers that are thousands of meters long.

2


Figure 3. NOAA’s Global Explorer [8], a remotely operated vehicle. Note the large yellow tether connecting the submersible to the operator.

The tether supports high data transfer rates and sometimes provides power to the vehicle. Because of the physical connection, tethers allow high data rate sensors, such as camera images and SONAR data, to be streamed in real time. The tether presents several challenges, however:

ROVs require a human operator to continuously supervise the vessel, whereas AUVs do not. [9]

Larger ROVs require a much larger support surface vessel, when compared to AUV operations, to carry the large reel holding the thousands of meters of tether cable. [10]

To support the weight and drag of the tether, deeper ROV operations require larger ROVs with larger thrusters to overcome the resistance. [11] [12]

Operating longer tethers with bigger ROVs requires more manpower to move the larger equipment. [10]

Conducting longer ROV operations with more personnel increases the overall cost.

AUV missions also suffer from the high costs of keeping operators at sea. A principal engineer in the Department of Applied Ocean Physics and Engineering from Woods Hole Oceanographic Institution, a private maritime research facility, estimated their cost of AUV operations: “Cost

depends on the AUV. Small R100 is cheap $500 to $1000 per day. Deep R6000 with Launch and Recovery and all support equipment $10,000/day.” [13].

Efforts to develop platforms that would reduce or eliminate the need for excessive manpower have led to the development of Autonomous Surface Vehicles (ASV). Besides requiring little human interaction, ASVs may be useful in AUV operations. When floating on the surface, ASVs can use wireless radio communications with the operators. Satellite communications between the shore and an ASV eliminate the need for operators at sea. The use of GPS is also available to surface vehicles. With this navigation system, ASVs can travel across large bodies of water without human intervention and still arrive within meters of the desired destination.

Although ASVs are effective autonomous research platforms, by themselves they are not very useful in underwater operations. Due to attenuation, imaging SONAR is limited to a range of several hundred meters. To be useful, an ASV with an imaging SONAR would be limited to shallow water; deep-sea searches would not be possible.

All three platforms—ASVs, AUVs, and ROVs—have strengths and weaknesses when performing underwater operations. A system designed to use the strengths of all three platforms would greatly improve AUV operations. An autonomous system would decrease the cost of operating the system to a fraction of the current operations cost by reducing the required man power. This paper presents one such solution.

THEORY OF THE SYSTEMOne way of combining the three systems would require placing the ROV onboard the ASV. Such a system would work like this:

1. The ASV travels to the location of the operating AUV.

2. Once on station, the ASV launches the ROV, and the ROV dives to a predetermined

3


depth. The depth could be as deep as the AUV or limited by the length of a shorter tether.

3. Eventually the AUV connects with the ROV and transfers its data.

4. Once data transfer is complete, the AUV disconnects from the ROV and continues operating without resurfacing.

5. The data from the AUV is sent to the ASV via the tether.

6. The data is transferred to the operator from the ASV.

7. The ROV is recovered when operations conclude.

Harnessing the strengths of each platform would create an effective underwater operations system.

The benefits from using this system are many. One benefit is the ability to download data from the AUV without requiring surfacing and recovery. Some AUVs have a diving speed of 45 meters/min and an ascent speed of 135 meters/min [14]. At this rate, it would take the AUV almost two hours to reach the estimated depth of MH370 at 4,500 meters [15] and more than 30 minutes to surface. Adding another two hours for AUV preparations, launching, and recovery brings the entire process close to four hours. During the four hours, the AUV could not perform its missions, and the time would be lost. To minimize this loss in efficiency, operators desire to limit the underwater time between AUV launch and recovery to maximize the operation’s efficiency. However, with the three-platform system, data from the AUV could be transferred to the user without requiring the AUV to surface. This allows AUV missions to be divided into smaller runs where the operator receives smaller amounts of data more often. The search missions would also be more effective because the operators can react and redirect the search in response to the real-time data.

The three-platform system also solves the cumbersome problem of submersible vehicle recovery. Surfacing AUVs can be difficult to locate especially in rough seas [16]. When surfaced, most AUVs are barely above the water

and can only be seen when operators are within several miles. If the AUV surfaces in the wrong location, such as underneath a ship or iceberg, the vessel could be damaged or even lost. Using the ROV to connect to the AUV, the AUV could be reeled out of the water, which would simplify recovery and safeguard the submersible. Docking underwater is advantageous because the ocean is much calmer beneath the surface. Engineers have designed a system that uses an ASV-dragged, drogue-based mechanism to recover the AUV [17]. A drogue-based recovery system also benefits from the calm conditions of underwater docking. However, because the ASV must pull the drogue to maintain depth, the method is not effective in areas that may restrict vessel maneuverability, such as narrow channels or deep crevices.

In addition to wireless communications, the ASV also provides controllable mobility. Engineers designed a similar idea in using ROVs with AUVs that involved a buoy launched from an aircraft that used a tethered ROV [18]. The ROV from the buoy would dive to the desired depth and attach with another submersible. However, the buoy system was subject to any ocean currents or winds. The buoy was unable to maintain position and would pull the ROV away from the operating area.

Using a different approach, much research has explored the development of anchored buoys that provide docking mechanisms, data transfer, and even charging [19] [20] [21] [22]. However, these buoys are fixed and would be very difficult to implement in deep waters. If the AUV operations began to move away from the anchored buoy, the increased time required to travel between the AUV and buoy would eventually negate the benefits of underwater recovery. The anchored system had a limited operating depth. In very deep water, the cable length required to keep the buoy anchored would be prohibitive.

In contrast, an ASV, using its thrusters, could maintain its position or follow the operations of the AUVs. The ASV system would not be depth-dependent and could overcome mild ocean currents and winds. To prove the above

4


concepts, students at the University of Florida Machine Intelligence Laboratory (MIL) are developing a prototype called the NaviGator Autonomous Maritime System (AMS).

PROOF OF CONCEPTThe objective of the NaviGator AMS is to validate the above proof of concept in a real environment. MIL students are developing three vessels:

NaviGator, the ASV Anglerfish, the ROV SubjuGator, the AUV

All three platforms will operate in freshwater and ocean environments. The proof of concept will be considered successful when data from SubjuGator is transferred to NaviGator ASV through Anglerfish.

NAVIGATOR ASVNaviGator ASV, shown in Figure 4, is a Wave Adaptive Modular Vessel (WAM-V) outfitted with a computer, sensors, and thrusters. The WAM-V consists of two inflatable pontoons connected by a suspension system and a large platform for the mounting of sensors, computers, and other equipment. The suspension system dampens the vessel’s oscillations, caused by high sea states, and prevents the rear-mounted thrusters from leaving the water making propulsion more efficient by allowing the hull to conform to the waves [23] [24]. The suspension system also attempts to keep the platform horizontal. Table 1 provides an overview of the vessel.

Figure 4. The NaviGator ASV operating at Lake Wauburg, Gainesville FL. NaviGator ASV uses four trolling motors to propel it through the water.

Mass 295 kgLength 5.5 mTop Speed 3 kt (1.5 m/s)Propulsion type Brushed trolling

motorsThruster force Four thrusters, 353 N

eachPower supply Two 24 V lithium ion

batteriesOperating time 5 hoursHull material Aluminum structure

on inflatable pontoons

Table 1. Specifications of NaviGator, the ASV.

For propulsion, NaviGator uses four trolling motors that each produces 353 N of thrust shown in Figure 5. The motors are mounted in fixed orientations that allow movement in all three degrees of freedom available to surface vessels (sway, surge, and yaw). Two 24 V lithium ion batteries, with 104 Ah of electric charge, provide approximately five hours of operation. NaviGator has a maximum range of 15 nautical miles.

5


Figure 5. A CAD model of the NaviGator ASV. The four trolling motors can be seen below the two pontoons.

A custom-built Intel I7-based computer with the Ubuntu Linux operating system and custom software utilizes the onboard sensors and controls the thrusters. A student-designed Integrated Navigation System (INS) estimates NaviGator ASV’s roll, pitch, and yaw. The INS utilizes accelerometers, gyros, magnetometers, and GPS to estimate the vessel’s position and orientation (POSE). Cameras and a LIDAR provide the system with data that can be used to avoid obstacles. A large radio antenna provides wireless communications between the boat and the operator. NaviGator ASV can carry and launch the ROV from beneath the main platform.

ANGLERFISH ROVAnglerfish [25], shown in Figure 6, is designed to be controlled by the NaviGator ASV. A tether is connected between the ROV and the ASV that provides power to the ROV and communication between the two. The 30 m tether is strong enough to be used as a method to recover the submersible. Table 2 gives the specifications of Anglerfish.

Figure 6. The Anglerfish ROV with eight thrusters visible.

Mass 7 kgTop Speed 1 kt (0.5m/s)Propulsion type Brushless motor

thrustersThruster force Eight thrusters, 20 N

eachPower supply Powered through

tetherOperating time 5 hours Max depth 100 mHull material Acrylic tube with 3D

printed structureTable 2. Specifications of the Anglerfish ROV.

An acrylic watertight tube contains the electronics. A 3D printed structure is designed around the tube to accommodate the mounting of the thrusters, lights, and actuators. Eight thrusters are mounted in a configuration that allows movement in all six degrees: roll, pitch, yaw, heave (depth), surge, and sway. The thruster positions can be seen more clearly in Figure 7.

6


Figure 7. A top-down view of Anglerfish. Note the four vertical thrusters to help with roll, pitch, and heave.

A Raspberry Pi 3 development board running Ubuntu Linux manages the thrusters and monitors the sensors. Anglerfish’s Inertial Measurement Unit (IMU) estimates the submersible’s roll, pitch, and yaw. The development board also monitors a pressure sensor that determines the submarine’s depth. Visual feedback to the Anglerfish ROV is provided through a camera mounted inside the dome. A small, wireless router is installed inside the water-tight tube and placed at the front of the ROV. This router is used to establish a short-range wireless network link with the SubjuGator AUV.

Students considered several ideas for the docking process between Anglerfish and SubjuGator. A popular AUV docking method requires specific body geometry for docking. This method utilizes a cone-like cage that a torpedo-shaped AUV can physically enter [26] [27]. This mechanism limits the types of AUVs that can be docked. To accommodate SubjuGator AUV’s box-like frame, students designed an alternative docking mechanism.

Docking with SubjuGator is achieved with the capturing system mounted underneath the dome of Anglerfish. The system consists of two electromagnets and four LEDS shown in Figure 8. The LEDs provide a visual reference for SubjuGator during the docking process. The LED orientation is inspired by research into space docking techniques for underwater

docking [28]. During wireless data transfer, the electromagnets hold the two submersibles together. Once the data transfer is complete, the vehicles will disengage.

Figure 8. The docking mechanism on Anglerfish. Four blue LEDs are used by SubjuGator as a visual reference during the docking procedure.

SUBJUGATOR AUVStudents selected SubjuGator as the prototype AUV because of its modularity, maneuverability, and underwater operating time. Figure 9 shows the front of the vessel. Modularity is achieved by the carbon fiber frame that permits the easy mounting of additional components. One such device is the steel docking point for Anglerfish, which is discussed later. Table 3 gives the specifications of SubjuGator.

Figure 9. SubjuGator, the fully autonomous submarine. The carbon fiber tubular frame provides multiple mounting locations for external devices.

7


Mass 43 kgTop Speed 1 kt (0.5 m/s)Propulsion type Brushless motor

thrustersThruster force Eight thrusters, 88 N

eachPower supply Two 24 V lithium

polymer batteriesOperating time 1.5 hoursMax depth 150 mHull material Aluminum hull with

carbon fiber structureTable 3. Specifications of the SubjuGator AUV.

SubjuGator consists of two waterproof aluminum hulls mounted on an aluminum plate and held together with a carbon fiber frame. The primary hull contains the computer and power management circuitry. The navigational equipment is located inside the secondary hull. Eight thrusters are mounted in an orientation that allows the AUV to move in all six degrees of freedom as shown in Figure 10. Two 24 V batteries, connected in series, support system operation for 90 minutes.

Figure 10. A top-down view of SubjuGator showing its thruster configuration. Each thruster is marked with a yellow circle.

Using the Ubuntu Linux operating system and custom software, the server-grade, Intel Xenon processor-based computer in the submarine’s primary hull controls the vessel. An IMU provides the roll, pitch, and yaw of SubjuGator.

The Doppler Velocity Log (DVL) sensor, shown in Figure 11, is mounted to the bottom of the submarine. The sensor provides an estimate of the vehicle’s three linear velocities. Depth is determined by a pressure sensor. Four hydrophones are used as part of the passive SONAR. Finally, the two cameras in the front of the submarine provide stereo vision feedback that is used during the docking procedure.

Figure 11. This image shows the bottom of the DVL mounted on SubjGator. The red discs use ultrasonic signals to determine SubjuGator’s velocity in the water.

Mounted on the front, horizontal beam of SubjuGator is the docking point shown in Figure 12. The device consists of a 3D printed mount and a steel plate. The steel plate is used as the holding point for the ROV’s electromagnets. A Wi-Fi adapter is installed inside the 3D printed mount. The adapter is used to wirelessly connect to the router on Anglerfish.

Figure 12. A close up of the mounting point on SubjuGator. The wireless adapter is located behind the steel plate.

8


SPECIALIZED HARDWARE AND SOFTWAREAll three systems utilize Ubuntu Linux and the Robot Operating System (ROS). ROS is a software framework that is specific to and greatly enhances the coding of robotic software [29]. ROS allows data, in the form of messages, to be shared across a network among different programs and platforms. Any program can publish or receive messages. This ability forms a cloud-like network that allows for decentralized operations among multiple systems. High volumes of data can be sent from a less powerful computer to a more powerful computer where it can be analyzed more rapidly, a function that is important in utilizing the Anglerfish ROV as an extension of the boat. Camera and other sensor data can be transferred from the ROV to the boat through the tether. NaviGator can then analyze the data and issue a command to the ROV based on the received data and current mission.

Wi-Fi was selected as the communication medium because it works at short ranges underwater and does not require physical contact. 2.4 GHz Wi-Fi has been proven to work reliably at ranges under 13 cm [30] [31]. Figure 13 shows the results from a study of transmission distances of 2.4 GHz underwater wireless communications. Because of underwater wireless range, Wi-Fi communications require less positional accuracy than physical connections. When Anglerfish can connect to SubjuGator’s mounting block, wireless communications can be established. The estimated distance between the router on Anglerfish and the adapter on SubjuGator, when the two are docked, is approximately 10 cm.

Figure 13. A graph showing received packets versus distance between antennas of underwater 2.4 GHz transmissions [30].

Anglerfish’s sensor suite provides roll, pitch, yaw, and depth (heave), but the sensors cannot provide surge and sway. Estimates of all six degrees are necessary to control the ROV accurately. On more sophisticated submersibles, surge and sway can be measured by taking the double integral of the accelerometer readings. Unfortunately, Anglerfish uses affordable yet somewhat noisy accelerometers, which prohibits the use of taking the double integral of the accelerometer data for surge and sway calculations. Another sensor that could be used to measure surge and sway is a DVL such as the one mounted on SubjuGator AUV.

DVLs are very expensive (SubjuGator’s DVL cost $25K), and mounting the sensor would require drastic changes to the mechanical structure of Anglerfish. Therefore, another solution was implemented. Inspired by indoor quadcopter controls using external cameras [32], students developed a method to measure the ROV’s velocity using an external sensor. In clear water, a submerged, downward-looking camera mounted on NaviGator ASV’s hull can be used to track Anglerfish’s relative movements. The software for tracking Anglerfish visually is run on NaviGator’s more powerful computer. The output from the tracking software (surge, sway) is sent to Anglerfish through the tether as shown in Figure 14. NaviGator ASV’s downward-looking camera is shown in Figure 15. Adding a bright green LED to the submersible improves the

9


camera’s ability to track the ROV’s movements. Figure 16 shows the LED on Anglerfish.

Figure 14. The above diagram shows the flow of sensor feedback. Anglerfish needs six degrees of feedback. Two are provided by NaviGator.

Figure 15. NaviGator ASV’s submerged, downward-looking camera. The camera is mounted inside a waterproof case.

Green was selected because of its low absorption in water [33]. Green was also selected because it would not interfere with the blue docking lights on Anglerfish. The observed movements from NaviGator ASV’s camera are combined with the sensor data provided by Anglerfish’s onboard sensors to create an accurate estimate of the ROV position. However, because the visual tracking depends on water clarity, this solution may not be reliable and will be used only until an acoustic-based tracking system (presently under development in MIL) is available.

Figure 16. The white dome is the LED bank. It will emit a green light that NaviGator ASV’s submerged camera will track.

The acoustic-based system consists of an active transducer and a passive SONAR system on NaviGator ASV. Anglerfish is outfitted with an acoustic transducer that periodically emits an acoustic signal (ping). NaviGator ASV uses a student-designed passive SONAR system, shown in Figure 17, that monitors signals from four hydrophones to determine the bearing of the acoustic transmissions [34].

Figure 17. The passive SONAR system. The four black hydrophones and the 3D printed mount are in the foreground. The student-designed SONAR circuit board is in the background.

The bearing estimated from the SONAR on NaviGator ASV is combined with Anglerfish’s reported depth. This combination provides an accurate three-dimensional position of the ROV, relative to the ASV. The acoustic system can operate within an estimated 45 meter maximum

10


distance between the two vessels. When compared to the camera-based method, the sound-based system provides a longer range and does not depend on water clarity.

Both NaviGator ASV and SubjuGator are outfitted with the same type of passive SONAR, and both listen to the ROV’s transmissions. However, instead of passively tracking the ROV, SubjuGator uses its SONAR to approach the ROV from a long distance during the homing process. SubjuGator estimates the ROV’s bearing and depth using a two-dimensional fix (angle and declination).

AUTONOMOUS SYSTEM CONTROL For docking to be successful, all three systems must be controllable and stable. NaviGator and SubjuGator have undergone hundreds of hours of in-water testing. Both have controllers that work well as shown in the following graphs. However, Anglerfish, with less than five hours of testing, still has a rudimentary controller. Further testing is required to improve the controllability and stability of the ROV. For the purposes of this paper, surge is movement along the x-axis, sway is movement along the y-axis, and heave is movement along the z-axis.

NaviGator ASV uses a model-reference adaptive control (MRAC) controller. This controller uses a pre-defined model of the vessel to improve tracking and can compensate for state-dependent disturbances such as constant winds or surface currents.

Figures 18 and 19 show NaviGator’s orientation and positional error during testing in a lake. The boat was station-holding in light rain with approximately 5 m/s of wind. Under these conditions, NaviGator maintained its heading with a yaw standard deviation (SD) error of 0.005 radians. The ASV also held its position well with an x-position SD error of 0.035 meters and y-position SD error of 0.028 meters.

Figure 18. NaviGator ASV was able to hold its heading very accurately during light weather.

Figure 19. Very little positional error was recorded during testing with the ASV.

NaviGator can maintain its heading accurately because of its controller and INS. The INS consists of a custom circuit board that uses an extended Kalman filter to fuse data from the GPS receiver, accelerometers, gyros, and magnetometers. This system has been proven to be very accurate with a positional drift of only several centimeters in one hour.

SubjuGator AUV uses a Robust Integral of the Sign of the Error (RISE) controller [35]. This is

11


another controller that can compensate for bounded disturbances and uncertainties of a dynamic system. An advantage of this controller is its ability to track without the oscillations experienced by other controllers such as a proportional derivative (PD) controller.

In Figures 20 and 21, the orientation and positional errors are shown during the AUV’s station-holding at a depth of two meters in a testing pool. SubjuGator was able to hold its orientation with a roll SD error of 0.006 radians, a pitch SD error of 0.008 radians, and a yaw SD error of 0.007 radians. Its position deviation is equally impressive with an x-position SD error of 0.005 meters, a y-position SD error of 0.003 meters, and a z-position SD error of 0.003 meters.

Figure 20. This graph shows that SubjuGator held its orientation well, with pitch having the highest SD at 0.008 radians.

Figure 21. Data provided by the DVL on SubjuGator provided very good positional feedback to the RISE controller. The feedback allowed SubjuGator to maintain its position within centimeters.

While the controller on SubjuGator was implemented and tuned well, the controller was only as good as the orientation and position feedback it was provided. On SubjuGator, accurate feedback is provided by the platform’s INS. The submarine’s INS consists of an IMU, water pressure sensor, and DVL. A Kalman filter fuses the sensors together to estimate the POSE of the ASV. The DVL is critical in estimating position. Without the DVL, determining position is very difficult.

A PD controller was used to control Anglerfish. This type of controller uses much less processing power when compared to a RISE or MRAC controller. The lower processing power demand is important because the controller must be run on the ROV to reduce latency. The processor on Anglerfish is inferior when compared to the computers on SubjuGator and NaviGator.

The PD controller on Anglerfish has been tested in a pool several times. Figures 22 and 23 show how Anglerfish held its orientation and position while station keeping in 0.5 meters of water. Anglerfish held its orientation with a roll SD error of 0.165 radians, pitch SD error of 0.113

12


radians, and a yaw SD error of 0.334 radians. The ROV’s z-position was maintained with a SD error of 0.070 meters. During testing, the camera tracking system had not been implemented yet, so only the depth of the ROV was recorded.

Figure 22. This graph shows that Anglerfish’s PD controller can hold the vessel’s orientation with oscillations. The +-1.0 radian yaw errors were caused by purposely induced disturbances from a diver.

0 25 50 75 100-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25Anglerfish Position Error

Z-Position Error

Time (s)

Pos

ition

Erro

r (m

)

Figure 23. Z-Position error varied by less than +-0.2 m. Further testing will help reduce the error.

While some stability has been achieved, the ROV suffers from oscillations. This may be due to the uneven ballast of the system or an improperly tuned controller. Further testing will be performed to better tune the controller. Next, an integral term will be added to dampen the system and reduce the SD errors.

Anglerfish may never be as controllable as SubjuGator, for several reasons. First, the PD controller is not as adaptable as the RISE controller. PD controllers struggle with disturbances. Second, Anglerfish is using an inexpensive IMU and does not have a DVL. This is the main reason why Anglerfish will struggle with its position estimate. To compensate during docking, the more controllable vessel (SubjuGator) will adjust to the less stable vessel. SubjuGator will move to dock with a stationary Anglerfish.

DOCKING AND DATA TRANSFERThis application controls docking using electromagnets and a steel plate because the mechanism does not need a high degree of accuracy. The magnet only needs proximity to the metal plate to connect. Mechanical locking mechanisms, such as a probe and socket or grabbing claw, require much greater accuracy. Another advantage of the magnetic design is safety. If the system fails, it will fail open and release SubjuGator. Two electromagnets provide a higher probability that one will connect with the plate when compared to a single magnet system. When magnetically connected, the ASV will be close enough to transfer data wirelessly. Once the transfer is complete, SubjuGator will be magnetically released.

The following steps outline the three-vessel operation from deployment to docking to data transfer. Before the operation can begin, all three platforms must be deployed and on station:

1. NaviGator ASV is deployed and maneuvers to a given point.

2. Once on station, the ASV holds its position and heading.

13


3. Anglerfish is then launched from the ASV, begins transmitting once wet, and descends to a desired depth.

4. When the ROV reaches the desired depth, it maintains its position and heading.

5. SubjuGator is then deployed within SONAR detection range of the ROV.

6. The AUV descends to roughly the same depth as Anglerfish or the required depth of the AUV’s mission.

When the AUV’s mission is complete, the homing and docking processes begin:

1. The AUV begins listening for the acoustic transmitter. A bearing estimate is made when the transmitted acoustic signal is received.

2. Using the good bearing estimate, the AUV begins maneuvering toward the direction of the transmissions.

3. SubjuGator updates Anglerfish’s perceived bearing with each new received transmission and continues moving towards the emitter.

4. During this process, SubjuGator constantly searches for the reference LEDs on Anglerfish to begin proper docking alignment.

5. Once SubjuGator visually recognizes the LEDs on Anglerfish, the AUV shifts to a visual docking mode.

6. The AUV maneuvers into Anglerfish using the lights as a visual reference as shown in Figure 24.

7. When the AUV is near the ROV, the electromagnets on Anglerfish pull and hold both vessels together, completing the homing operation.

Figure 24. Before transferring data, Anglerfish connects to the steel mounting point on SubjuGator using two electromagnets.

The final process is the data transfer:

1. After the ROV connects to the mounting point, SubjuGator attempts to connect to the wireless router on Anglerfish.

2. Establishing a wireless connection allows SubjuGator to transfer files to Anglerfish.

3. The files are then relayed to the ASV. 4. Once the transfer is complete,

SubjuGator disconnects from the ROV, concluding the exchange.

Operators on shore maintain contact with all three platforms. However, the ASV and ROV are on separate networks from the AUV. Data from SubjuGator is transferred to NaviGator ASV’s network. Once data from SubjuGator is transferred to NaviGator ASV’s network, the above process, currently a proof of concept, will be considered proven.

FUTURE WORKDeploying more than one AUV for use with the ASV/ROV system would further improve the efficiency of AUV operations and would not require additional man power. Multiple AUVs could divide the work and efficiently cover a larger search area as a combined system. Each AUV could then be summoned individually using an assigned acoustic signal. As the flotilla of submersibles advanced with their search areas, the ASV and ROV could follow along, reducing the distance each AUV would need to travel during the homing period. This system

14


would greatly improve operational efficiency during operations such as mapping or searching the ocean floor.

Placing an AUV recharging system on the ASV would allow the vessel to charge an AUV through the ROV. Assuming the ASV could recharge itself with a renewable energy source such as solar energy, and this could be shared with AUVs, the AUVs could stay submerged and operate indefinitely. Non-contact recharging methods would allow more flexibility in the recharging process and could accommodate a variety of AUV hull shapes. [36] [37].

CONCLUSIONThis paper presents a three-vessel system for underwater operations. Combining the strengths of the tethered ROV with the utility of the AUV improves underwater operations by creating a high-speed underwater communications path. Docking the AUV with the ROV also eliminates the need for the AUV to surface in order to transfer sensor data to the operators. When the above method is implemented through the use of an unmanned ASV, the high costs associated with long AUV operations can be greatly reduced by eliminating the need for personnel at sea.

REFERENCES

[1] "Missing flight MH370: Robotic submarine to begin search," 14 April 2014. [Online]. Available: http://www.bbc.com/news/world-asia-27017928. [Accessed 14 July 2016].

[2] "Bluefin-21 Gallery," Bluefin Robotics, [Online]. Available: http://www.bluefinrobotics.com/assets/Galleries/Bluefin-21/Bluefin-21-Testing.jpg. [Accessed 19 July 2016].

[3] "Search for MH370 Facts and statistics Surface search of the southern Indian Ocean 17 March – 28 April 2014," 28 April 2014. [Online]. Available: http://jacc.gov.au/search/files/MH370_Facts_and_statistics-Surface_search_of_the_southern_Indian_Ocean.pdf. [Accessed 15 July 2016].

[4] J. Linshi, "Time," This Is the Country That’s Spent the Most Searching for MH370, 11 June 2014. [5] L. Lanbo, Z. Shengli and C. Jun-Hong, "Prospects and problems of wireless communications for

underwater sensor networks," Wireless Communications and Mobile computing, pp. 977-994, 1 October 2008.

[6] "DP900-OIL Series Sonar User Handbook," January 2016. [Online]. Available: http://www.blueview.com/assets/Uploads/downloads/p-series-deep-oil-203087-01-P900-OIL-Series-REV-A.pdf.

[7] M. S. Lawrence, "Synthetic Aperture Sonar Survey to Locate Archaeological Resources in the Stellwagen Bank National Marine Sanctuary," National Oceanic and Atmospheric Administration, 26 August 2010. [Online]. Available: http://oceanexplorer.noaa.gov/explorations/10sbnms/welcome.html. [Accessed 19 July 2016].

[8] "The Hidden Ocean, Artic 2005," National Oceanic Atmospheric Administration, 27 June 2005. [Online]. Available: http://oceanexplorer.noaa.gov/explorations/05arctic/logs/summary/media/global_explorer_rov.html. [Accessed 19 July 2016].

[9] R. D. Christ and R. L. Wernli, "Standard operational procedures," in The ROV Manual: A User Guide for Remotely Operated Vehicles, Waltham, Butterword-Heinemann, 2014, pp. 602-623.

[10] R. D. Christ and R. L. Wernli, "LARS and TMS," in The ROV Manual: A User Guide for Remotely Operated Vehicles, Waltham, Butterworth-Heinemann, 2014, pp. 221-248.

[11] R. D. Christ and R. L. Wernli, "ROV Size Classifications," in The ROV Manual: A User Guide for Remotely Operated Vehicles, Waltham, Butterworth-Heinemann, 2014, pp. 5-8.

[12] R. D. Christ and R. L. Wernli, "Drag Discussion," in The ROV Manual: A User Guide for Remotely Operated Vehicles, Waltham, Butterworth-Heinemann, 2014, pp. 73-88.

15


[13] M. Purcell, Interviewee, Woods Hole Principal Engineer in the Department of Applied Ocean Physics and Engineering. [Interview]. 7 July 2016.

[14] A. MacNaughton and R. P. Swanson, "A new navy deep-water oceanographic survey vehicle: the Naval Oceanographic Office's Subsurface Autonomous Mapping System (SAMS)," in OCEANS 2003, San Diego, 2003.

[15] H. Yan and E. Lavandera, "How deep is deep? Imagining the MH370 search underwater," CNN, 25 April 2014.

[16] K. Hardy and B. Rosenthal, "Surface Detection For Recovery of UUVs," Sea Technology, pp. 43-45, March 2015.

[17] E. Sarda, M. Dhanak and K. V. Ellenrieder, "Concept for a USV-Based Autonomous Launch and Recovery System," in Launch and Recovery 2014, ASNE, Linthicum, 2014.

[18] S.-C. Yu, K. Y. Yi, T.-W. Kim, S. K. Choi, T. Ura and J. Yuh, "Autonomous Undersea Vehicle Applications Center," [Online]. Available: http://auvac.org/uploads/configuration_spec_sheets/sauvim.pdf. [Accessed 28 March 2016].

[19] D. E. Frye, J. Kemp, W. Paul and D. Peters, "Mooring developments for autonomous ocean-sampling networks," IEEE Journal of Oceanic Engineering, vol. 26, no. 4, pp. 477 - 486, 2001.

[20] H. Singh, J. G. Bellingham, F. Hover, S. Lemer, B. A. Moran, K. von der Heydt and D. Yoerger, "Docking for an autonomous ocean sampling network," IEEE Journal of Oceanic Engineering, vol. 26, no. 4, pp. 498-514, 2001.

[21] T. Fukasawa, T. Noguchi, T. Kawasaki and M. Baino, ""MARINE BIRD", a new experimental AUV with underwater docking and recharging system," in OCEANS 2003, San Diego, 2003.

[22] R. Stokey, N. Purcell, N. Forrester, T. Austin, R. Goldborough, B. Allen and C. v. Alt, "A Docking System for REMUS, an Autonomous Underwater," in OCEANS '97, Halifax, 1997.

[23] A. W. Peterson and M. Ahmadian, "Simulation and Testing of Wave-adaptive Modular Vessels," in 11th International Conference on Fast Sea Transportation, Honolulu, 2011.

[24] A. Peterson, M. Ahmadian, M. Craft and A. Shen, "Simulation and scale-testing to improve the next generation of wave adaptive modular vessels," in 2013 Grand Challenges on Modeling and Simulation conference, Toronto, 2013.

[25] A. Gray and E. Schwartz, "Anglerfish: an ASV controlled ROV," in The 29th Florida Conference on Recent Advances in Robotics, Miami, 2016.

[26] B. W. Hobson, R. S. McEwen, J. Erickson, T. Hoover, L. McBride, F. Shane and J. G. Bellingham, "The development and ocean testing of an AUV docking station for a 21" AUV," OCEANS 2007, IEEE, pp. 1-6, 2007.

[27] S. Knepper, M. Niemeyer, R. Galletti, A. Brighenti, A. Bjerrum and N. Anderson, "EUROCKER - A Universal Docking - Downloading - Recharging System for AUVs," WIT Transactions on The Built Environment, vol. 53, pp. 427- 435, 2001.

[28] M. Francesco, Y. Petillot, A. Mallios, S. Krupinski, R. Haraksim and P. Sotiropoulos, "Investigation of portability of space docking techniques for autonomous underwater docking," in OCEANS 2009 - EUROPE, Bremen, 2009.

[29] M. Quigley, B. Gerkey, K. Conley, J. Faust, T. Foote, J. Leibs, E. Berger, R. Wheeler and A. Ng, "ROS: an open-source Robot Operating System," in ICRA workshop on open source software, Kobe, 2009.

[30] S. Sendra, J. Lloret, P. C. Rodrigues and J. M. Aguiar, "Underwater Wireless Communications in Freshwater at 2.4 GHz," IEEE Communications Letters, vol. 17, no. 9, pp. 1794 - 1797, 2013.

[31] S. Sendra, J. V. Lamparero, J. Lloret and M. Ardid, "Underwater Communications in Wireless Sensor Networks using WLAN at 2.4 GHz," in 2011 IEEE Eigth International Conference on Mobile Ad-Hoc and Sensor Systems, Valencia, 2011.

[32] L. C. Mak and T. Furukawa, "A 6 DoF visual tracking system for a miniature helicopter," in 2nd

16


International Conference on Sensing Technology, Palmerston North, 2007. [33] W. S. Pegau, D. Gray and R. V. Zaneveld, "Absorption and attenuation of visible and near-infrared

light in water: dependence on temperature and salinity," Applied Optics, vol. 36, no. 24, pp. 6035 - 6046, 1997.

[34] J. Easterling and E. M. Schwartz, "Pocket Passive SONAR," in The 29th Florida Conference on Recent Advances in Robotics, Miami, 2016.

[35] N. Fischer, H. Devin, P. Walters, E. Schwartz and W. Dixon, "Nonlinear RISE-based control of an autonomous underwater vehicle," IEEE Transactions on Robotics, vol. 30, no. 4, pp. 845-52, 2014.

[36] T. McGinns, C. P. Henze and K. Conroy, "Inductive Power System for Autonomous Underwater Vehicles," in Oceans 2007, Vancouver, 2007.

[37] T. Kojiya, F. Sato, H. Matsuki and T. Sato, "Construction of non-contacting power feeding system to underwater vehicle utilizing electro magnetic induction," in Europe Oceans 2005, Brest, 2005.

[38] J. A. Simpson, B. L. Hughes and J. F. Muth, "Smart Transmitters and receivers for underwater free-space optical communication," IEEE Journal on selected areas in communications, vol. 30, no. 5, pp. 964-974, 2012.

[39] M. Waheed-uz-Zaman and M. Yousufzai, "Design and Construction of Very Low Frequency Antenna," Journal of Basic and Applied Sciences, vol. 7, no. 2, pp. 141-145, 2011.

[40] J. A. Simpson, W. C. Cox, J. R. Krier, B. Cochenour, B. L. Hughes and J. F. Muth, "5 Mbps optical wireless communication with error correction coding for underwater sensor nodes," in Oceans 2010 MTS/IEEE Seattle, Seattle, 2010.

[41] R. T. Howard, T. C. Bryan and M. L. Book, "An advanced sensor for automated docking," in Digital Avionics Systems, 2001, Daytona Beach, 2001.

[42] S.-C. Yu, J. Yuh and J. Kim, "Armless underwater manipulation using a small deployable agent vehicle connected by a smart cable," Ocean Engineering, vol. 70, pp. 149 - 159, 2013.

[43] "Acoustic Modems," Teledyne Benthos, [Online]. Available: http://teledynebenthos.com/product_dashboard/acoustic_modems. [Accessed 19 July 2016].

[44] "Debris Belongs to Doomed Malaysia Airlines Flight MH370, Experts Say," Yahoo News, 25 March 2016. [Online]. Available: https://www.yahoo.com/news/debris-belongs-doomed-malaysia-airlines-flight-mh370-experts-133548602.html?ref=gs. [Accessed 19 July 2016].

[45] "ASV Successfully Demonstrates C-Worker Unmanned Surface Vehicle in USA," Unmanned Systems Technology, 17 April 2014. [Online]. Available: http://www.unmannedsystemstechnology.com/2014/04/asv-successfully-demonstrates-c-worker-unmanned-surface-vehicle-in-usa/. [Accessed 19 July 2016].

ACKNOWLEDGEMENTSI acknowledge the tremendous support of my advisors at the University of Florida’s Herbert Wertheim College of Engineering: Dr. Eric Schwartz, Professor Carl Crane, Professor Antonio Arroyo, the Machine Intelligence Laboratory (MIL), the Center for Intelligent Machines and Robots (CIMAR), the Department of Electrical and Computer Engineering, and the Herbert Wertheim College of Engineering, all at the University of Florida.

AUTHOR BIOGRAPHIESANDREW C. GRAY, MS is a graduate student in Electrical and Computer Engineering at the University of Florida. He received his BS degree in Systems Engineering from the United States Naval Academy and served five years in the United States Navy as a naval officer on two warships. In 2014, he

17


received his MS degree in Electrical and Computer Engineering from the University of Florida and is presently a PhD candidate.

ERIC M. SCHWARTZ, MS, PhD (mentor) is a Master Lecturer in the Department of Electrical and Computer Engineering at the University of Florida and the Director of the Machine Intelligence Laboratory. He has BS degrees in both Electrical and Mechanical Engineering and an MS and PhD in Electrical Engineering, all from the University of Florida.

18

€¦ · web viewin addition to wireless communications, the asv also provides controllable...

Documents