Collision and occlusion free positioning algorithms for ROV-manipulator systems

In the last years the interest concerning Underwater Robotics has significantly grown. The need of inspecting pipes, perform cable maintenance or monitoring the healthiness of offshore structure's in an human hostile environment has highlighted the importance of developing vehicles to remotely accomplish these tasks. Currently it is common the use of manned robotic systems, however this approach presents some strong limitations due to the enormous cost and risk of working at the bottom of the sea for a human operator. Furthermore, the technology nowadays involved in the context of underwater manipulation typically makes use of a master/slave approach. In this framework, a skilled operator is required to move the master arm which acts as a joystick for the slave manipulator that is performing the task. Clearly this strategy presents several limitations such as the need of a well-trained operator, the significant delay between the master and slave arm as well as the fact that underwater communication is challenging overall. This thesis is meant to be a first step towards the development of autonomous Underwater Vehicle Manipulator Systems (UVMS), in particular we are going to set the basis for the implementation of coordinated control strategies between the manipulator and the underlying robotic structure. Specifically, we will develop positioning algorithms (Inverse Kinematics) for a manipulator equipped onto a Remotely operated Underwater Vehicle (ROV) aiming at maintaining a desired posture despite of the undesired motion of the robotic structure due, for instance, by water currents. In the following, therefore, several techniques for solving the positioning task in compliance with the mechanical limits of the robotic arm are investigated and tested. Moreover, a planner based on the Bidirectional-RRT (BiRRT) algorithm capable of generating occlusion and collision free motions for the manipulator is presented and implemented . This is of fundamental importance since for the sake of checking whether the task is being correctly performed or not. Finally, the results are tested using the Unity environment in order to validate the results on a more reliable basis and have a pictorial intuition of how the algorithms behave.