A SURVEY ON UNINHABITED UNDERWATER VEHICLES (UUV)

More documents

Recommendations

Info

its environment. In [29], the authors designed a Multiinput/ Multi-output self tuning control system and theperformance of the system is evaluated.An adaptive Saturated Proportional-Derivative (SP-D) setpoint controller for AUVs is proposed in [30]. It isclaimed that the global asymptotic stability can beguaranteed even when the gravity and buoyancy force isuncertain with this type of a controller. In this work,Lyapunov’s direct method and LaSalle’s invarianceprinciple are used.A hybrid adaptive control of AUV is investigated in[31]. The authors presented the results for modified direct,indirect, and linear quadratic Gaussian adaptive control inthis study. Results concluded that the direct methodproduced the best results. In [32], the authors introduced avehicle control system that is capable of learning andadapting to changes in the vehicle dynamics andparameters. This control system is compared with aconventional linear control system in this study.In [33], the authors described a technique to localizethe vehicle through detecting known landmarks using onboardnavigation sensors, overcoming uncertainties invehicle dynamics and its operating environment.Simulations and experiments conducted in this studyindicate that the system is capable of controlling thevehicle in six DOF with high accuracy using the estimatedposition and velocity from the on-board sensor-basednavigation system. In [34], the authors presented a sixdegrees-of-freedom controller for AUVs. The authorselected the control law to be adaptive to cope with thedynamic parameters which are mostly uncertain in theunderwater environment. The proposed control law adoptsquaternion representation for attitude errors, and thusavoids representation singularities that occur when usingEuler angles description of the orientation.An adaptive control of underwater robots with sonarbasedposition measurements is presented in [35]. In thelight of the test results, the control system is shown not tonot require any prior information about the systemdynamics and yet could provide high performance in thepresence of noise and unmodelled dynamics. In [36],another adaptive control law for AUVs and ROVs isintroduced which is not using earth-fixed frame norvehicle-fixed frame. Results indicate that the use of theproposed adaptive action could significantly reduce thetracking error.In [37], the authors proposed two methods tocompensate for the model uncertainties. The first methodis an adaptive passivity-based control scheme and secondis a hybrid (adaptive and sliding) controller. In this work,the hybrid controller is simulated for the horizontalmotion of the Norwegian Experimental RemotelyOperated Vehicle (NEROV) and simulation resultsindicate a satisfactory performance. In [38], the authorsinvestigated the theory and experimental work of theAdaptive plus Disturbance Observer (ADOB) controllerfor underwater robots. They claim that this controllerwould be robust under the influence of externaldisturbance and uncertainties in the system. The result ofexperiments conducted in this study indicate that theADOB controller could be listed as promising forunderwater robots, especially for the systems with failingPID type controllers.Yuh has presented the results of recent study on theapplication of neural network to the underwater roboticvehicle system in [39]. The robustness of the system wasinvestigated against the nonlinear dynamic behavior andthe results proposed by computer simulations. Accordingto the simulation results using neural network hasincreased the autonomy of the vehicle. In [40], the authorsintroduced learning control approach to underwaterrobotic vehicle system using neural networks. Thesimulation tests result concluded that dominant vehicledynamics are varying with the vehicle velocity and effectof thruster dynamics becomes significant at low velocityin the vehicle control system. Yuh [41], has also presenteda learning control system which is using neural networksfor underwater robotic vehicles and the system is tested bysimulations. The simulation test results indicate thatcontrol system is capable of providing an acceptabletracking performance. A direct adaptive neural networkcontrol system is designed in [42]. The researchers trainedthis sytsem on-line by parallel recursive error predictionmethod and critic equation. In [43], the authors alsodescribed a neural network system which can arrange therobot dynamics and controller adaptation in parallel withrobot control. In this work, adaptability of the system isinvestigated under unknown disturbance.In [44], a neuro-fuzzy controller for autonomousunderwater vehicles (AUVs) is described. The authorsclaimed that the advantage of modified fuzzy membershipfunction-based neural networks (FMFNN) is combiningthe fuzzy logic and neural networks. Compared to othercontrol methods, the proposed FMFNN control algorithmsaid to never require any information on systems, off-linelearning procedures, and human intervention to adjustparameters. Computer simulations for this work areconducted and result indicated that the proposed FMFNNcontroller for an unknown dynamic system producesacceptable performance compared to other real-time, selftuningcontrollers. In [45], the authors present theutilization of a self adaptive neuro-fuzzy controller as afeedforward controller and in mean time PD control as afeedback controller in controlling an AUV. In [46], a newdesign for fuzzy logic controller is proposed. In the studychanges in UUV depth, regulates pitch are simulated andobserved.UNDERWATER VEHICLE-MANIPULATORSYSTEM (UVMS)The control of UVMSs can be investigated as puremotion control and interaction control architectures. In35.4
order to deploy a force-reflecting bilateral teleoperationsystem, interaction control architecture is our main area ofinterest.Dynamic Motion Control of UVMSsThe problem of redundancy resolution and motioncoordination between the vehicle and the manipulator inUVMSs are addressed in [47]. In this study, a task-priorityinverse kinematics approach to redundancy resolution ismerged with a fuzzy technique in coordination of thevehicle-arm. Researcher conducted simulation studies ona 9-dofs UVMS. In [48], a robust control scheme using amultilayer neural network with the error back propagationlearning algorithm is proposed. The proposed controller isemployed in the control of a robot manipulator operatingunder the sea which has large uncertainties such as thebuoyancy, the drag force, wave effects, currents, and theadded mass/moment of inertia. Simulation test resultsindicated that the control scheme could with theunexpected large uncertainties. In [49], the adaptivepassivity-based control scheme is formulated in anaugmented task-space where both the underwater vehicleand the end-effector have 6 degrees of freedom.A tele-robotic control system has been developed fora subsea manipulator as part of the ARM (AutomatedRemote Manipulation) project in [50]. The ability of theARM System is that it conducts tasks, previously thoughtto require divers, allows for the greater use of ROVs incurrent fields with considerable economic savings. In[51], the authors presented a new method to analyzedynamics of underwater robot manipulators. In theproposed method, hydrodynamic terms such as addedmass, drag and buoyancy in dynamics of underwaterrobots are obtained by iterative learning control and timescaletransformation. In [52], the authors developedhighly-accurate model of the hydrodynamic interactionforces, and implemented a coordinated arm/vehiclecontrol strategy. Under this model-based approach,interaction forces acting on the vehicle due to arm motionwere predicted and fed forward into the vehicle controlsystem. Using this method, vehicle station-keepingcapability was greatly enhanced. Tracking errors andsettling times for the manipulator end point were reducedsignificantly.Interaction Control of UVMSsTwo control schemes, extended hybrid control andextended impedance control, for compliant motion controlof redundant manipulators in [53]. The experimentalresults have validated the two control schemes anddemonstrated that the redundancy can be effectivelyutilized to optimize various objective functions. In [54],method of actively controlling the apparent stiffness of amanipulator end effector was presented. The approachallows the programmer to specify the three translationaland three rotational stiffness of a frame located arbitrarilyin hand coordinates. The stiffness control approach toforce control in a manipulator system has been shown tobe a useful and effective means of effecting force controlin assembly tasks.In [55], the authors presented a force control strategyfor a robot floating on the water. The control strategyreduces the number of vehicle actuators required for theforce control by utilizing the restoring force/momentapplied to the vehicle. In [56], the authors defined aproper metric in joint space. Minimal parameterization ofmotion and force controlled subspaces as wel1 as the nullmotion component is realized. With this formulation,control of both motion/force and internal motion ofredundant manipulator could be achieved utilizing a newhybrid impedance control method with inertial decouplingof each space. In [57], the authors presented a spatialimpedance control with redundancy resolution. In order toensure geometric task consistency, the rotational part ofthe stiffness is described in terms of a unit quaternion. Thedynamically consistent pseudoinverse of the manipulatorJacobian is adopted to decouple the dynamics of the endeffectormotion from the null-space motion. Redundancyis exploited to stabilize null-space joint velocities andoptimize an additional task function.LOCALIZATION/NAVIGATIONMost conventional systems have to use a GPS systemto localization purposes where they have to go up to thesurface after some period of navigation. Some recentresearch activities have focused on developing newsensory systems and strategies to cancel this type of aprocedure. The following are some examples to theseefforts.At Florida Atlantic University, the researchers [58]designed and developed an enhanced inertial navigationsystem that is to be integrated into the Morpheusautonomous underwater vehicle. A complementary filterwas implemented to provide a much smoother and stableattitude estimate. The results show that the filteringperformance can be considered acceptable when the errorautocorrelation function falls within a tolerable limit. In[59], the authors introduced a new methodology for thedesign of multi-rate navigation systems for underwatervehicles. The design technique proposed borrows fromKalman filtering theory and leads naturally to multi-ratecomplementary filtering structures, the performance ofwhich can be assessed using a frequency-like domaininterpretation.In [60], the authors combined a database whichcontains sonargrammetric, terrain matching, and imageregistration information with the standard navigation withthe standard navigation instrument suite. The accuracy ofpositional estimates could be maintained over a longerduration. As a result of this adaptive calibration, it isdiscussed that it is no longer necessary to go up to thesurface to get new position from GPS. In [61], the authors35.5
Page 1 and 2: ASME Early Career Technical Journal
Page 6 and 7: provided a navigation method of an
Page 8: [42] Lorentz, J. and Yuh, J., 1996,

A SURVEY ON UNINHABITED UNDERWATER VEHICLES (UUV)

Create successful ePaper yourself

Delete template?

Save as template?