This paper describes a model-based thruster fault detection scheme for an autonomous underwater vehicle. For a vehicle equipped with servo motor based marine thrusters, the velocity and current feedback from the motor controllers can be used to derive two independent thrust approximations. The difference between the two models reveals the presence of fault conditions, and quantifies the error of the output thrust relative to the desired reference. The method is proven effective through experimental validation.

This paper presents the theory and experimental work of the adaptive plus disturbance observer (DOB) controller on underwater robot. This control scheme consists of DOB as an inner-loop compensator and a non-regressor based adaptive controller as an outer-loop controller. It is regressor-free, and is robust with respect to external disturbance as well as the variation of the nominal model used to construct DOB. The performance and characteristics of this controller are shown through comparison tests among four different controllers: PID controller, PID plus DOB controller, adaptive controller and adaptive plus DOB controller.

This paper identifies the key issues in the empirical implementation of redundant marine thrusters for two different autonomous underwater vehicles of different designs, so as to maximize the effectiveness for fine motion control. A crucial factor in the development of a precision control scheme is the thorough understanding of the system geometry. Other critical elements include the contribution of the thruster amplifier gains to the steady state response of the thrusters, and the influence of the power source on system performance. These findings are proven through experimental analysis and verification.

During the last decade, the first autonomous underwater robot of the Autonomous Systems Laboratory of the University of Hawaii, ODIN (Omni-Directional Intelligent Navigator) built in 1991 has produced a lot of valuable results in development and control methods [(S.K. Choi et al., July 1994), (Choi, S.K. et al., March 1995), (S.K. Choi et al., April 1996), (Yang, K.C. et al., September 1999), (J. Yuh et al., 2000)]. Recently, ODIN was born again in the 3rd generation with unique features under recent technologies such as abundant system resources owing to a PC104+, new vehicle system software architectures with an object-oriented concept and its implementation, a graphical user interface and an independent algorithm module using a dynamic linking library (DLL) based on the Windows operating system. These give us an ideal environment for developing various algorithms which are needed for developing an advanced underwater robotic vehicle. This paper describes details of the development of ODIN-III and presents initial experimental results for fine motion control.

This paper considers thruster dead zones and saturation limits in fine motion control of underwater robots. Thruster dead zones and saturation limits are major nonlinear elements that make fine motion control of underwater robots difficult. If the vehicle is designed with a redundant thruster configuration, the dead zone and the nonlinear region around it can be completely avoided by implementing a null motion solution for the command input of the vehicle. This solution is derived primarily from the hardware geometry, and is actually realized before the application of the motion control algorithm. The result is an improvement in system performance regardless of the implemented controller type. This paper describes the mathematical derivation of a null motion solution, the minimum energy solution, and a hybrid of the two. The approach is then illustrated through simulation, and followed by empirical application to an underwater vehicle that was developed at the Autonomous Systems Laboratory of the Mechanical Engineering Department of the University of Hawai'i at Manoa.

This paper describes the development of a laser-based ranging device, as part of a sensor work package that is to be integrated into an autonomous underwater vehicle. The objective of the laser ranging component includes robust operation in turbid ocean environments from 1 to 10 meters with a balance of accuracy and speed. A calibration technique that utilizes a spline interpolation is employed as an accurate, yet simple alternative to conventional triangulation methods. The resulting configuration is capable of range detection within 2% absolute error, at a 10 Hz refresh rate, out to 4 meters in turbid ocean conditions, and to greater than 10 meters in air. The maximum underwater range of the system is currently limited by the existing hardware, whereas the software is ready for any future system upgrades.

The distributed virtual environment collaborative simulator is used for testing of unmanned underwater vehicles (UUVs) where both real and simulated vehicles can interact and cooperate in a hybrid, synthetic, virtual environment. This virtual reality system can be used to determine: 1) the optimal performance and criteria for the cooperating vehicles and its relative application; 2) the determination of the advantages and disadvantages of collaborative application tasks between multiple UUVs; and 3) the optimal communication links between the cooperating vehicles and its remote control stations. It also utilizes a virtual reality projection and polarized eye-wear systems, which immerses the user for optimal visualization.

This paper describes a distributed virtual environment collaborative simulator (DVECS) for testing of unmanned underwater vehicles (UUV) where both real and simulated vehicles-remotely operated vehicles (ROV) and autonomous underwater vehicles (AUV)-can interact and cooperate in a hybrid, synthetic, virtual environment containing various types of real and simulated vehicles, obstacles, conditions and disturbances. This virtual system can be used to determine (1) the optimal performance criteria for the vehicles and its application, (2) the advantages and disadvantages of collaborative application tasks between multiple UUV, and (3) the optimal communication links between remote locations. The current DVECS is being tested between the University of Hawaii in Honolulu and the University of Tokyo in Tokyo as a joint, international research program. It also utilizes virtual reality projection system, which immerses the user for optimal visualization.

Describes the design and implementation of a fault tolerant system for the 6 degrees-of-freedom ODIN autonomous underwater vehicle (AUV). Fault tolerant system design can be divided into three areas: fault detection, fault isolation, and fault accommodation. This paper summarizes previous studies in fault tolerant system design and presents a fault tolerant system design for thruster (actuator) and sensor failure in AUVs. The presented system was implemented in the actual vehicle, ODIN, and its effectiveness was evaluated with experimental results.

Many underwater robotic vehicles use propeller-type thrusters driven by DC motors. The thruster system is known to be nonlinear and time-varying. Robust control of the thruster system is crucial to navigate and achieve high performance. This paper presents experimental results of five different thruster control systems based on online neural network (NN) control, off-line NN control, fuzzy control, adaptive-learning control and PID control. These control systems were initially designed or trained for the thruster system with no fin. Then each controller was implemented for the thruster system modified with different size fins. The fins changed the system hydrodynamics, especially the drag force. In this paper, each control system is briefly described and its practical advantages and disadvantages are discussed with experimental results.

Underwater robotic vehicles (URVs) have become an important tool for various underwater tasks because of their greater speed, endurance, depth capability and a lower risk factor than human divers. However, most vehicle control system designs are based on simplified vehicle models and often result in poor vehicle performance due to the nonlinear and time-varying vehicle dynamics having parameter uncertainties. This paper describes a new vehicle control system capable of learning and adapting to changes in the vehicle dynamics and parameters. This control system is compared with a conventional linear control system through extensive wet tests. Results show the learning and adapting capabilities of the presented control system.