Model-based Trajectory Generation and Control of Continuum Manipulators

• Applicant: Professor Dr.-Ing. Oliver Sawodny
• Subject Area: Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics
• Term: since 2016
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 317742086

The aim of the project is the development of model-based control algorithms for redundant continuum robots. The developed methods are to be investigated using pneumatically actuated continuum manipulators as example systems. These offer a high potential in the environment of cooperation with a human being. Typically, human-machine interactions are characterized by two basic tasks: the detection of external forces and the realization of a given behavior in the case of contact. For this purpose, the existing model will first be extended and examined in greater detail. Based on this, model-based approaches for the detection of the contact case at the manipulator will be investigated: How can contact situations be reliably detected and to what extent is it possible to precisely estimate the point of application, direction, and magnitude of an applied force from the specific properties of the continuum manipulator and the given pressure and angle sensors. Due to the design of the manipulator, the stiffness of the bellows segments can be specified via the mean pressure in addition to feedback control. This mean pressure is to be taken into account in a central unified control, which serves as a preparation for the task planning.In addition to the typically considered trajectory of the tool center point, the contact with the environment and the associated force are also taken into account in the task planning. During the planning itself, the question arises how the additional degrees of freedom of the pneumatically actuated continuum manipulator can be used to primarily ensure the feasibility of the defined task and, secondarily, to implement the task itself in a state that is as advantageous as possible for the ma-nipulator, such as stiffness and maximum force that can be applied, but also system-theoretical properties, such as favorable observability with regard to the contact situation. Finally, interactive algorithms are to be developed for the task definition, which allow a human to provide the manipu-lator with a path and a force reference by haptic deformation, which can then be implemented using the developed planning algorithms.

• DFG Programme: Research Grants