Underwater robotic vehicles have been very attractive for scientific survey, construction, and maintenance in hazardous environment such as ocean and nuclear plant. The development of advanced vehicle technologies is accelerated by the demand for an underwater robotic vehicle with high performance and efficient operation in terms of time and cost. Most vehicle control systems based on the simplified vehicle model often fail to provide desired performance because of the vehicle's nonlinear and time-varying dynamics in nature. This paper describes two vehicle controllers: nonlinear controller and adaptive controller considering the effect of thruster dynamics. Results of the controllers tested for the pitch and altitude motion are presented.

The authors present the application of discrete-time adaptive control to an underwater robotic vehicle (URV) having a mechanical manipulator. During the working mode, the coordinated control system compensates for errors due the manipulator motion and provides high performance in terms of speed and accuracy. Therefore, by using the control approach presented, one can achieve the efficient operation of the vehicle and its manipulator for the desired performance. The control strategy presented does not require prior knowledge about the vehicle system parameters.>

An approach to the robust position and velocity control of mechanical manipulators having a joint compliance is presented. A control model of such a manipulator is derived by linearizing the nonlinear equations about a desired trajectory. This control model is described by multivariable difference equations. The parameters of the model are estimated by parameter-adaptation algorithms. Control gains, determined from the control model by using the pole-placement technique, are computed online with the estimates of the system parameters. Simulation results with a two-link mechanical manipulator are presented to show the effectiveness of the approach.>

An algorithm generating the dynamic equations of the manipulator in a symbolic form is presented. The proposed algorithm is based on a modified Lagrange-Christoffel formulation of robot dynamics. A PC-based program, ARDEG, implements the algorithm and automatically generates the equations of motion for open-chain manipulators in a symbolic form. Various types of manipulators are investigated to evaluate the efficiency of the program. In a 3-DOF PUMA robot, the ARDEG-generated algorithm reduces the computational load of a general Newton-Euler recursive algorithm by 80% and that of a recursive algorithm generated by the computer program ARM by more than 22%.>

A dynamic model of the link, which has rotational and translation motion, is derived using a Lagrangian equation. The validity of the model is investigated by comparing the results of computer simulation to experimental results for a single-link arm. For the experiment, a two-degree-of-freedom flexible robot arm was built, and tip deflections of the arm were observed by double integrating tip-acceleration measurements.>