Bipolar half plates (BPPs) are important parts of Proton Exchange Membrane (PEM) hydrogen fuel cells. Metallic BPPs are generally produced by stamping or hydro forming. However, these processes are not continuous and relatively slow compared to rotary forming. Rotary forming is a continuous process and could potentially be a faster BPP production method. The BPPs are plastically deformed in bending and membrane modes. It is necessary to determine the bending and membrane strain because differentiating them can aid in the design of future BPP rotary forming production methods. Bending strain predominantly could lead to fractures on the outside of the bend. Membrane strain causes material thinning and could potentially lead to necking. These deformation modes need to be well understood so that the highest forming levels are achieved with a high level of geometrical consistency without exceeding material failure limits. The goal of this research is to determine the bending and membrane strain of 0.1 mm thick SS316L sheet material formed by rotary forming. This is done by experimental testing and analysis. From the experimentally deformed material, cross-sections are made and analysed. The thickness and bending strain are measured from these cross-sections. The experimental results are compared to FEA simulations. An important assumption is plane strain and it is experimentally validated. Plane strain FEA models are used to further understand the problem. The FEA simulations are validated, using the single channel strain experiments. Simulations with channels aligned with the cylinder axis are validated with a maximum thickness error of 3%. However, channels perpendicular to the axis are validated with an error of 8.9%. The FEA simulations are then used to further analyse equivalent plastic strain, stress, springback and final shape. The maximum equivalent plastic strain is 0.34. The maximum VonMises stress is 1200 MPa and maximum springback is 0.075 mm. The final shapes along the cylinder axis meet the experimental results the best. The bending and membrane strain are experimentally measured and simulated in Abaqus CAE. The bending and membrane strain highly depend on channel orientation and indentation depth. However, the bending strain is greater than membrane strain for both experiments and simulations. This research should be extended to investigate more complex geometries and designs. This is of course a more realistic example for BPPs.

Task-parameterized movement representation, as an approach for the generalization of demonstrations, is used to represent data from multiple local perspectives within the global reference frame, through which more accurate information about multiple aspects of the movement is given. The estimated transformation between the different perspectives and the global reference frame in task parameter inference can be used for gesture recognition. In this thesis, task parameter inference in the application of human-robot interaction, a method called TP-inference approach, is investigated. It consists of a combination of task parameter inference and task parameter movement retrieval. A task-driven model is used to generalize the demonstration data and the task parameter inference is achieved by using the orthogonal Procrustes analysis. The TP-inference approach is tested for various static tasks and is compared to the Probabilistic Movement Primitive (ProMP) approach [1]. The test results indicate that for simple and or similar movement of the human and robot, the TP-inference approach performs less accurate than the ProMP. For complex movements the TP-inference preforms more accurate than the ProMP. [1] M. Ewerton, G. Neumann, R. Lioutikov, H. B. Amor, J. Peters, and G. Maeda, Learning multiple collaborative tasks with a mixture of interaction primitives, in 2015 IEEE International Conference on Robotics and Automation (ICRA) (IEEE, 2015) pp. 1535–1542

Haptic Shared Control (HSC) can improve operator performance during teleoperation. HSC uses assistive forces on the master device to guide the operator's control input towards a reference trajectory. However, the reference trajectory might be incorrect due to inaccuracies in the sensed or modeled environment and thus not match the intended trajectory of the operator, resulting in force conflicts that can potentially reduce performance. This paper describes the design and evaluation of a haptic shared controller whose reference trajectory adapts in real-time to resort force conflicts. Different adaptive controllers have been designed and compared in simulation. The final controller is built on a Task Parameterized Gaussian Mixture Model and adapts using an Incremental Gaussian Mixture Model. The controller is evaluated in an experiment (n=16) where subjects performed a planar teleoperation task that involved moving the slave through a narrow corridor. The subjects were assisted by HSC with a correct reference (cHSC), HSC with a wrong reference (wHSC), HSC with a wrong reference that adapts in real-time (waHSC) and a manual condition. The waHSC condition was expected to result in lower control effort, compared to wHSC, and induce better performance compared to wHSC and the manual condition. Within-subjects results showed that waHSC decreases control effort compared to wHSC, while the performance remained similar. Manual control had better performance than wHSC and waHSC, and cHSC outperformed all other conditions. Even though the conditions were tested in a balanced order, we observed a strong effect of the order of the conditions: between-subject results showed improved performance, lower control effort and better subjective results for waHSC over wHSC when wHSC was executed first. Upon first execution of waHSC, improved performance and self-reported satisfaction were realized for wHSC over waHSC. The results show that, in the event of a force conflict, an adaptive controller is a promising alternative to temporarily adjusting the HSC stiffness.