MOTION CONTROL. During months M25 to M30, ICCS further progressed towards the development of methods that enable motion planning and control of the robotic manipulator in a workspace with obstacles. The objective is for the robot to move near the body area and to follow a predefined bathing behaviour on it. The obstacles considered in this work may in particular regard restricted areas on the user's body, which should be avoided during the washing sequence (for instance, as specified in a particular user profile). These areas may be located on the body part on which the robot operates, but may also concern other body parts that interfere to the robot's motion (e.g. the hands of the user). The core task of motion planning is to calculate online the reference pose for the end-effector of the robotic arm, which will let the soft arm to execute surface tasks (e.g. scrubbing the user's back) and at the same time to be compliant with this body part. This is a challenging task, since all human's body parts are non-planar surfaces, that are moving and deform (e.g. user's breathing motion) or stochastically. Moreover, the user either systematically will have constant communication with the system using audio-gestural commands and will be able to change the state and the parameters of the system with motions, which may interfere to the robot's operation. Adaptability to specific user needs and requirements is also considered as a very important property of the system. Hence, the basic course is to plan the motion of the robot on a 2D “canonical” space, which is normalized in terms of duration of the motion and size of the space in each dimension. In particular, proper washing motions for each subtask are learned by demonstration from professional carers and nurses. Besides the use of Dynamic Motion Primitives (DMP) approaches, work in this reporting period has further progressed in the development of a motion planning method that uses ideas and algorithms based on a Navigation Functions approach, by means of generating motions through repulsive potential fields(“virtual obstacles”) that are reconstructed and learned from demonstration data, as shown in Figure 19.
MOTION CONTROL. In the motion control user interface the user can interact with the robot to control the motion. The user could use a joystick to control the motion of the robot in a 3D space (target) 🡺 left / right / up / down He also can steer the speed of the robot: rotating speed of the drill head and pushing speed of the propulsion mechanism. Figure 10: Control display of a HDD drill rig TRACTO-TECHNIK
MOTION CONTROL. The highly nonlinear nature of the vehicle dynamics, the presence of strong currents and turbulences as well as the need – in case of coordinated motion – to precisely coordinate the motion of multiple vehicles renders the problem of motion control of underwater vehicles a very challenging task. Although there has been an intensive effort, especially recently, towards developing efficient motion control methodologies for underwater vehicles, see e.g. [M1]–[M9], the problem of accurate and efficient motion control designs for underwater vehicles still remains an open issue. One should note that if exogenous disturbances, e.g. strong currents and turbulences were perfectly known (i.e. measurable and predictable), the problem of accurate and efficient underwater vehicle motion control would have found a satisfactory solution even when coordination between multiple vehicles is required. Unfortunately, the existing – successful in the known- disturbance cases – methods may perform very poorly, or even fail when strong currents or turbulences affect the vehicle. Attempts to “robustify” the aforementioned methods lead to very
