Robotics is increasingly attracting interest for more and more applications. In particular, underwater robotics is having a large development since the need of mineral resources is growing. The latter pushes to find out new deposits on the seabed. Therefore, the construction and the maintenance of underwater structures are necessary, e.g., submarine pipelines for carrying oil and gas. However, performing any task on the seabed is very dangerous for the man, for obvious reasons. Thus, several Autonomous Underwater Vehicles (AUVs), equipped with manipulators as well, have been implemented in recent years, aimed at physically substituting the man. The control of Underwater Vehicles-Manipulator Systems (UVMSs) require full control of the vehicle. Indeed, cruise vehicles with rudder and stern are not suitable for carrying a manipulator since they are not able to counteract the interaction forces with the arm itself. Furthermore, for a rigid body moving in a fluid there exist several hydrodynamic effects acting on it. In particular, among the latter, the restoring generalized forces, which are gravity and buoyancy, and the ocean current are of major concern in designing the control law since they influence the steady-state position and orientation errors. Beyond Proportional-Integral-Derivative actions (PID), several adaptive control laws have been proposed in literature for compensating these effects. However, they all are designed starting from the dynamic models written either in the earth-fixed or in the vehicle-fixed frame, respectively. Nevertheless, some hydrodynamic terms are constant in earth-fixed frame, e.g., the restoring linear force, and some others are constant in the vehicle-fixed frame, e.g., the restoring moment. Thus, in this thesis work, a mixed earth/vehicle-fixed frame-based adaptive control able to build each dynamic compensation action in the proper reference frame is proposed. In particular, a reduced version has been derived within the aim to achieve null steady state error under modelling uncertainty and presence of ocean current with respect to a minimal number of parameters. Furthermore, the effects of including the thruster dynamics within the full-dimensional adaptive control are investigated. Simulations and comparisons with other control laws, such as PID, show the better performance of the proposed technique. The proposed adaptive control law has been also tested within the EC-funded ROBUST Project with the interuniversity Center of Integrated Systems for the Marine Environment (ISME). The ROBUST system has been designed and implemented for performing sea bed material identification merging the capabilities of an AUV and a robotic manipulator with a LIBS (Laser Induced Breakdown Spectroscopy) sensor mounted on its end-effector. Thus, the adaptive technique has been used for the vehicle dynamic control. Redundant systems can be exploited to perform multiple tasks simultaneously. For this purpose, the Multi-Task Priority (MTP) inverse kinematics algorithm can be used. As well known in literature, the latter is based on the Closed Loop Inverse Kinematics (CLIK) and allows to manage a prioritized hierarchy of equality-based tasks, which are control functions characterized by a specific desired value, e.g, position and orientation. In addition to the latter there exists another category of control functions which are named set-based tasks since their value can range between an upper and lower bound, respectively. One of the most common set-based tasks is for instance the obstacle avoidance. Indeed, the obstacle distance has to respect a lower bound (a minimum distance value). However, it can assume values greater than the latter. Within the aim to control systems taking into account safety-related tasks, it is necessary to manage both equality and set-based tasks. Thus, the Set-Based Task-Priority Inverse Kinematics (SBTPIK) is proposed in this thesis. In particular, simulations and experiments with fixed and mobile base manipulators show the effectiveness of the algorithm as well as its integration into an assistive control framework for Remotely Operated Vehicles (ROVs). The SBTPIK validity is also demonstrated through its application within the EC-funded DexROV Project with ISME. More in detail, a via-satellite remotely controlled UVMS has been developed to perform several kind of tasks such as oil and gas pipelines maintenance. Therefore, it has been necessary to manage multiple tasks ensuring the safety system. Thus, the SBTPIK has been applied to fulfill this objective. The SBTPIK is a local motion control algorithm which efficiently performs on redundant systems since it handles real-time changes in the environment. However, it is prone to local minimum as any local motion controllers. Motion planners, on the other hand, are global methods and they are able to take into consideration the same system constraints. Nevertheless, their implementation often requires sacrificing some of the constraints or the redundancy exploitation. For this reason, in this thesis an approach based on merging the global and local planners in an effort to preserve the features of both ones is proposed. In particular, the global planner is implemented as a sampling-based algorithm which works in the reduced-dimensionality of the robot work space applying the Cartesian constraints only. The output trajectory is then checked against the inverse kinematics algorithm verifying the fulfillment of the other task constraints. The SBTPIK is then used also in real-time to ensure a reactive behaviour. During the movement, the motion planner runs in background to adapt to changes in the environment, human presence or, in general, to continuously optimize the path. The proposed method has been simulated within the DexROV and ROBUST frameworks and experimentally validated in laboratory with a mockup represented by the Kinova Jaco2 7 DOFs (Degrees of Freedom) manipulator. The SBTPIK framework has been successfully used in assistive applications as well, aimed at allowing users with severe motion disabilities to perform manipulation tasks that may help in daily-life operations. Tests have been performed using the Kinova Jaco2 7 DOFs manipulator operated via a P300-based Brain Computer Interface (BCI). More in detail, the P300 paradigm is based on the P300 potential which is a component of the Event Related Potentials (ERPs), i.e., a fluctuation in the EEG generated by the electrophysiological response to a significant sensorial stimulus or event. In particular, the P300 consists of a positive shift in the EEG signal approximately 300-400ms after a task relevant stimulus. Thus, the user with motion disabilities can generate command through a proper P300-based Graphical Interface (GUI). It is worth noticing that the present thesis focuses on underwater robotics therefore the BCI topic is not discussed in this work.
Planning and Control of Underwater Vehicle-Manipulator Systems
DI VITO, Daniele
2020
Abstract
Robotics is increasingly attracting interest for more and more applications. In particular, underwater robotics is having a large development since the need of mineral resources is growing. The latter pushes to find out new deposits on the seabed. Therefore, the construction and the maintenance of underwater structures are necessary, e.g., submarine pipelines for carrying oil and gas. However, performing any task on the seabed is very dangerous for the man, for obvious reasons. Thus, several Autonomous Underwater Vehicles (AUVs), equipped with manipulators as well, have been implemented in recent years, aimed at physically substituting the man. The control of Underwater Vehicles-Manipulator Systems (UVMSs) require full control of the vehicle. Indeed, cruise vehicles with rudder and stern are not suitable for carrying a manipulator since they are not able to counteract the interaction forces with the arm itself. Furthermore, for a rigid body moving in a fluid there exist several hydrodynamic effects acting on it. In particular, among the latter, the restoring generalized forces, which are gravity and buoyancy, and the ocean current are of major concern in designing the control law since they influence the steady-state position and orientation errors. Beyond Proportional-Integral-Derivative actions (PID), several adaptive control laws have been proposed in literature for compensating these effects. However, they all are designed starting from the dynamic models written either in the earth-fixed or in the vehicle-fixed frame, respectively. Nevertheless, some hydrodynamic terms are constant in earth-fixed frame, e.g., the restoring linear force, and some others are constant in the vehicle-fixed frame, e.g., the restoring moment. Thus, in this thesis work, a mixed earth/vehicle-fixed frame-based adaptive control able to build each dynamic compensation action in the proper reference frame is proposed. In particular, a reduced version has been derived within the aim to achieve null steady state error under modelling uncertainty and presence of ocean current with respect to a minimal number of parameters. Furthermore, the effects of including the thruster dynamics within the full-dimensional adaptive control are investigated. Simulations and comparisons with other control laws, such as PID, show the better performance of the proposed technique. The proposed adaptive control law has been also tested within the EC-funded ROBUST Project with the interuniversity Center of Integrated Systems for the Marine Environment (ISME). The ROBUST system has been designed and implemented for performing sea bed material identification merging the capabilities of an AUV and a robotic manipulator with a LIBS (Laser Induced Breakdown Spectroscopy) sensor mounted on its end-effector. Thus, the adaptive technique has been used for the vehicle dynamic control. Redundant systems can be exploited to perform multiple tasks simultaneously. For this purpose, the Multi-Task Priority (MTP) inverse kinematics algorithm can be used. As well known in literature, the latter is based on the Closed Loop Inverse Kinematics (CLIK) and allows to manage a prioritized hierarchy of equality-based tasks, which are control functions characterized by a specific desired value, e.g, position and orientation. In addition to the latter there exists another category of control functions which are named set-based tasks since their value can range between an upper and lower bound, respectively. One of the most common set-based tasks is for instance the obstacle avoidance. Indeed, the obstacle distance has to respect a lower bound (a minimum distance value). However, it can assume values greater than the latter. Within the aim to control systems taking into account safety-related tasks, it is necessary to manage both equality and set-based tasks. Thus, the Set-Based Task-Priority Inverse Kinematics (SBTPIK) is proposed in this thesis. In particular, simulations and experiments with fixed and mobile base manipulators show the effectiveness of the algorithm as well as its integration into an assistive control framework for Remotely Operated Vehicles (ROVs). The SBTPIK validity is also demonstrated through its application within the EC-funded DexROV Project with ISME. More in detail, a via-satellite remotely controlled UVMS has been developed to perform several kind of tasks such as oil and gas pipelines maintenance. Therefore, it has been necessary to manage multiple tasks ensuring the safety system. Thus, the SBTPIK has been applied to fulfill this objective. The SBTPIK is a local motion control algorithm which efficiently performs on redundant systems since it handles real-time changes in the environment. However, it is prone to local minimum as any local motion controllers. Motion planners, on the other hand, are global methods and they are able to take into consideration the same system constraints. Nevertheless, their implementation often requires sacrificing some of the constraints or the redundancy exploitation. For this reason, in this thesis an approach based on merging the global and local planners in an effort to preserve the features of both ones is proposed. In particular, the global planner is implemented as a sampling-based algorithm which works in the reduced-dimensionality of the robot work space applying the Cartesian constraints only. The output trajectory is then checked against the inverse kinematics algorithm verifying the fulfillment of the other task constraints. The SBTPIK is then used also in real-time to ensure a reactive behaviour. During the movement, the motion planner runs in background to adapt to changes in the environment, human presence or, in general, to continuously optimize the path. The proposed method has been simulated within the DexROV and ROBUST frameworks and experimentally validated in laboratory with a mockup represented by the Kinova Jaco2 7 DOFs (Degrees of Freedom) manipulator. The SBTPIK framework has been successfully used in assistive applications as well, aimed at allowing users with severe motion disabilities to perform manipulation tasks that may help in daily-life operations. Tests have been performed using the Kinova Jaco2 7 DOFs manipulator operated via a P300-based Brain Computer Interface (BCI). More in detail, the P300 paradigm is based on the P300 potential which is a component of the Event Related Potentials (ERPs), i.e., a fluctuation in the EEG generated by the electrophysiological response to a significant sensorial stimulus or event. In particular, the P300 consists of a positive shift in the EEG signal approximately 300-400ms after a task relevant stimulus. Thus, the user with motion disabilities can generate command through a proper P300-based Graphical Interface (GUI). It is worth noticing that the present thesis focuses on underwater robotics therefore the BCI topic is not discussed in this work.File | Dimensione | Formato | |
---|---|---|---|
Tesi_dottorato_Unicas_32_DI_VITO_Daniele.pdf
accesso aperto
Dimensione
17.06 MB
Formato
Adobe PDF
|
17.06 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/168569
URN:NBN:IT:UNICAS-168569