Set-Based Inverse Kinematics control of an UVMS within the DexROV project Paolo Di Lillo, Daniele Di Vito, Gianluca Antonelli University of Cassino and Southern Lazio 03043 Cassino, Italy {pa.dilillo,d.divito,antonelli}@unicas.it Abstract—In this paper we describe the algorithm developed for the motion control of an UVMS (Underwater Vehicle- Manipulator System) within DexROV, a funded EC Horizon 2020 project that proposes to enable the far distance teleoperation of an UVMS via satellite communication. The main use case scenario is the maintenance of Oil&Gas underwater platforms. During subsea oil operations, the robotic system has to be able to navigate to the panel location and manipulate handles, rotate valves, plug and unplug connectors. The DexROV system has two operational modes: fixed-base and floating-base manipulation. In floating-base manipulation the system is not clamped to any structure and it can move both the manipulator and the vehicle at the same time in order to perform the needed tasks. The control relies on a Task Priority Inverse Kinematics framework that has been extended in order to handle both equality-based and set-based tasks. Simulations performed in the Gazebo simulator show the effectiveness of the proposed approach. I. I NTRODUCTION Underwater robotics is of great interest because it allows to perform tasks that are very dangerous for the man, e.g., oil and gas submarine pipelines maintenance [1], [2]. Underwater Vehicle-Manipulator Systems (UVMSs) can be divided essen- tially in two main categories: Remotely Operated Vehicles (ROVs) and Autonomous Underwater Vehicles (AUVs). ROVs are systems physically linked via theter, used both for power supply and communication, to a surface vessel or a submarine where a human operator has to control and send commands to the vehicle itself. In particular a high level of expertise is required to the human operator in controlling and manoeuvring the vehicle made more difficult by the theter inertia. Thus the man has a fundamental role because he is the control loop. AUVs are systems with onboard power modules avoiding the theter. The absence of a physical link makes these vehicles more suitable to survey missions. Furthermore they present unmanned control loops contrarily to the ROVS. Thus they are supposed to perform the task for which they are designed without the presence of the man in the loop. However AUVs present a very low power autonomy. Therefore they are not widely used for maintenance-manipulation operations espe- cially inside the Industry but, however, in several scenarios they can be used in combination with ROVs exploiting both ones properties [3]. This work was supported by the European Community through the projects EUMR (H2020-731103-2), ROBUST (H2020-690416) and DexROV (H2020- 635491). In this paper we present the algorithm developed for the motion control of an UVMS within the scope of the DexROV project [4] (see Fig. 1). Fig. 1. A view of the DexROV UVMS. DexROV is a funded EC Horizon 2020 project that pro- poses to implement novel operation strategies for underwater semi-autonomous interventions. Currently these operations are performed tele-operating ROVs (Remotely Operated Vehicles) from a support vessel, and require the presence of a significant amount of personnel needed for the effective accomplishment of the intervention. The main goal of DexROV is to move the supervision of the mission onshore, allowing the usage of a much smaller support vessel with a reduced crew, resulting in a major decrease of the costs of the entire operation. The control center is placed onshore, and it is connected to the support vessel via a satellite link. Given the time latency introduced by the communication channel, it is not possible for the pilots to directly teleoperate the system anymore, and it has to exhibit some semi-autonomous capabilities. The control relies on a multi-task priority algorithm that allows the operator to focus only on the operational tasks, while the safety-related and the optimization ones are autonomously handled by the system. The DexROV system has two operational modes: fixed-base and floating-base manipulation. In floating-base manipulation the system is not clamped to any structure and it can move both the manipulator and the vehicle at the same time in order to perform the needed tasks in a more dexterous way. In this configuration, the system has 10 controllable DOF (Degrees Of Freedom): 4DOF are related to the moving base (the 3 components of the linear velocity and the angular velocity along the z axis), and 6 DOF are the manipulator’s joints. This redundancy has to be exploited to coordinate the