The acceptance of artificial assistance systems for humans depends on their ability to adapt to the specific needs of the user. For the motion assistance of the human upper limb by an active exoskeleton, the collaborative execution of this movement represents a fundamental research question. In this thesis, human-exoskeleton collaboration methods are investigated in order to enable safe and comfortable assistance during free movements of the arm and positioning of a tool for interaction with objects. An individual structural integration of sensor technology is applied to detect the user's intention to move. The sum of the dynamic influences on the motion sequence is determined through the structure-integrated measurement of the torque of the seven joint axes of the manipulator. Based on kinematic and dynamic models of the serial manipulator, the interaction of the user with the exoskeleton is derived with a measurement uncertainty of < 7 %. For the interaction of the manipulator with the environment, the haptic perception of the human can be substituted by a structure-integrated force sensor. The high degree of customization of the sensor geometry motivates novel manufacturing processes. In the context of this work, the fundamentals for the additive manufacturing of metallic force sensors are investigated.
A systematic design of a collaborative exoskeleton is carried out under consideration of application-related requirements. The prototype consists of seven actuated degrees of freedom to support the shoulder, elbow and wrist. For mobile use, the design is being further developed into a variant with a high level of integration through decentralized sensor-actuator nodes. This provides a torque of 30 Nm and a rotation speed of 3.14 rad/s with a total weight of 8.5 kg including the battery. The presented lightweight exoskeleton has a high innovation potential for the investigation of physiological aspects in collaboration with humans due to the modularly expandable sensor system for physical and cognitive measured variables.
The environmental interaction of the guided tool is implemented by force-regulated drilling in a bone phantom. A stable system behaviour with a maximum of 30 % overshoot by compensation of non-modellable compliance is achieved by means of a hybrid force-position controller of the end effector. For the trajectory planning, approaches for a natural arm movement are presented. For the transformation into the joint space of the redundant manipulator, the user intention is considered in the formulation of an optimization problem. Additional methods prevent singular postures and enable real-time trajectory planing.
The target pose of the trajectory is defined in advance on the virtual object of the phantom. Through a new line matching registration procedure, the real object is transposed on the virtual with fiducial marks. This reduces the required duration by 30 % and the error by 12.5 % compared to conventional registration methods. The relative position between the tool and bone phantom is recorded by structure-integrated markers with a tracking camera. The relative joint angle of the upper limb, on the other hand, is determined with an inertial sensor system. Sensor data fusion is used to determine the absolute position in space of the sensors attached to the arm segments and the individual sensor nodes are aligned with each other using a dynamic motion sequence. The joint angles, calculated with a maximum deviation of 3 %, serve as input variables for a model-based impedance control. A movement input is specified externally and comfortable guidance of the arm is ensured by adapting the joint angle stiffness. The human cognitive intention to move the arm is derived by determining the electrical activity during muscle contraction. A fuzzy logic reduces the interference of noise on the electrical signals of the antagonistic muscle pairs and assigns a corresponding joint angular velocity to the muscle activity. With the results of the movement prescriptions by the collaborative exoskeleton, the high demand for innovative approaches of individual movement therapy and support for humans is addressed and thus a contribution to the demographic change is made. | English |