A unified dynamics-based motion planning algorithm for autonomous underwater vehicle-manipulator systems (UVMS)

Abstract

IntroductionAmong the underwater robotic systems that are currently available, remotely operated vehicles (ROVs) are the most commonly used underwater robotic systems. A ROV is an underwater vehicle that is controlled from a mother-ship by human operators. Sometimes a ROV is equipped with one or more robotic manipulators to perform underwater tasks. These robotic manipulators are also controlled by human operators from a remote site (e.g., mother-ship) and are known as tele-manipulators. Although the impact of ROVs with telemanipulators is significant, they suffer from high operating cost because of the need for a mother-ship and experienced crews, operator fatigue and high energy consumption because of the drag generated by the tether by which the ROV is connected to the ship. The performance of such a system is limited by the skills, coordination and endurance of the operators. Not only that, communication delays between the master and the slave site (i.e., the mother-ship and the ROV) can severely degrade the performance. In order to overcome some of the above-mentioned problems, autonomous underwater vehicles (AUVs) are developed. However, an AUV alone cannot interact with the environment. It requires autonomous robotic manipulator(s) attached to it so that the combined system can perform some useful underwater tasks that require physical contact with the environment. Such a system, where one or more arms are mounted on an AUV, is called an autonomous underwater vehicle-manipulator system (UVMS). One of the main research problems in underwater robotics is how to design an autonomous controller for a UVMS. Since there is no human operator involved in the control of a UVMS, the task planning has become an important aspect for smooth operation of such a system. Task planning implies the design of strategies for task execution. In other words, a task planning algorithm provides a set of desired (i.e., reference) trajectories for the position and force variables, which are used by the controller to execute a given task. Task planning can be divided into motion planning and force planning. In this research, we focus on the design of motion planning algorithms for a UVMS. The motion planning of a UVMS is a difficult problem because of several reasons. First, a UVMS is a kinematically redundant system. A kinematically redundant system is one which has more than 6 degrees-of-freedom (DOF) in a 3-D space.

Coupled and Decoupled Force/Motion Controllers for an Underwater Vehicle-Manipulator System

“…Using the dynamic model of the system in the optimal trajectory generation, suitable trajectories can be generated for both vehicle and manipulator, taking into account the interactions between the two subsystems. A detailed description of this type of optimal trajectory generator for UVMS is presented in [31]. Nevertheless, this is out of the scope of this paper and represents a topic of itself that the authors aim to explore in a future paper.…”

Hybrid Strategy-based Coordinate Controller for an Underwater Vehicle Manipulator System Using Nonlinear Disturbance Observer

“…in some circumstances, it is preferable to move the manipulator rather than move the vehicle). Thus, the damped least-squares IK planner with weighted matrix can be employed by adopting the following Jacobian matrix: 10,25 J…”

“…Besides, the switching time point is congruously defined at t = 50 s, the unknown lump term τ ζ d is exactly the same during different controllers. In other words, the tracking controller and the IK planner will create identical trajectories for three controllers before the switching point, and then the system will switch into the end-effector controllers as the adjustment phase which will last for 40 s. In simulations, we define ψ ηp = diag [15,15,20,10,10,10,10,10] ψ ηd = diag [15,20,10,12,12,12,112,12] ψ ζ p = diag [16,16,25,25,25,25,25,25,25,25] ψ ζ d = diag [10,10,12,12,12,12,12,12,12,12] K ηp = diag [8,8,16,16,16,16,16,16] K ηd = diag [8,8,…”

“…By employing the null space of the inverse Jacobian matrix and using the gradient descent method, 9 the IK planner technique developed rapidly from the velocity-level optimization to the acceleration-level optimization in the UVMS domain. [10][11][12][13] However, due to the numerical integrations and differentiations in the IK plan process, numerical errors are, however, inevitable in the planning routine on each DOF. It can lead to accumulated errors on the end-effector level.…”

Task-priority redundancy resolution on acceleration level for underwater vehicle-manipulator system

“…The GPM has been successfully applied to the optimization of fluid drag force, obstacle avoidance, and torque optimization. [6][7][8][9] In this method, a scalar potential function to be optimized is established firstly and then its negative gradient is projected into the inverse kinematics solution by means of the null space operator. In the study by Ji et al, 10 zero moment point equation is chosen as the function to be optimized.…”