Framework for Autonomous Underwater Grasping of Unknown Objects

Abstract

This thesis studies the problem of autonomously grasping an unknown object in an underwater scenario using an I-AUV. Firstly, a new approach to autonomously obtain an accurate and complete 3D reconstruction of an object of interest is presented. Next, it is detailed an algorithm to control redundant kinematics having into account hard constraints. Then, an algorithm to calculate the pose of the gripper that fulfill, in a grater way, a hierarchy of tasks and that strictly accomplishes a set of constraints. The following proposed methodology guides the gripper to the planned pose avoiding collisions. Finally, two methodologies for avoiding problems during the intervention and fix them are explained. The first controls the position and orientation of the arm end-effector. The second detects contacts between the gripper and the environment, and correct the end-effector trajectory to avoid them.