This study proposes a novel design of monolithic compliant variable torsional stiffness (VTS) mechanism whose stiffness can be tuned continuously. This mechanism is then used to create a compliant variable stiffness (VS) ball joint. This compliant VS ball joint can be beneficial in various applications such as exoskeletons, prostheses, and bio-mimicking due to their high adaptability to environmental changes and the capability for multitasking.

Compliant mechanisms are a popular alternative to conventional mechanisms because of their advantages on mass reduction, friction, backlash and lubrication. However, the major drawback in the use of compliant mechanisms is the stiffness is the desired direction of motion. The advantages of conventional and compliant mechanisms can be combined if the stiffness can be reduced or ideally removed, resulting in a zero-stiffness compliant mechanism. In current designs of zero-stiffness mechanisms, a preload is applied in the stiffest (compression) direction of a flexible beam. This working principle is based on Euler buckling. In this work, compliant mechanisms with zero-stiffness behaviour are obtained using lateral torsional buckling as their working principle. For this purpose, two compliant joints are modelled in a FEA, a translational and a rotational joint. The results of the FEA are verified by experiments. Also, a sensitivity analysis is performed on the cross-sectional dimensions of the flexible beams to optimize the range of zero-stiffness. Both types of joints showed zero-stiffness behaviour for the optimal preload and a good agreement is found between simulations and experiments. From the sensitivity analysis is found that a rectangular cross-section results in the largest range of zero-stiffness.

This project presents a compliant slender mechanism with a low axial-bending stiffness ratio. This can be used in passive exoskeletons. Passive exoskeletons can assist people with their work by preventing injuries. Currently exoskeletons uses sliders to account for the extension of the spine when bending, but this could be uncomfortable. The proposed mechanism tries to solve this problem by having a low axial stiffness while maintaining its bending stiffness. It consist of vertical rigid bodies connected by horizontal flexures. The distance between the flexures is found to be a variable almost independent of the axial stiffness, thus able to easily influence the ratio. To help with future designs a PRBM-model is proposed that describes the proposed mechanism. This is then used in a dimensional optimization algorithm. The found design is then produced as a prototype and tested. The axial-bending stiffness ratio that has been found was 8.5 and the maximum error of the PRBM model 18%.

In this work, the possibilities to approach various nonlinear moment-angle characteristics with a kinematically indeterminate rigid body balancer with torsion springs are examined. These torsion springs are mounted on the axes that intersect the rigid bodies. The rigid body balancer is coupled to an inverted pendulum. Although the kinematic indeterminate nature of the system enables the balancer to rotate non-proportionally along with the pendulum, the kinematics should correspond with the equilibrium configurations of the system. The required system parameters as spring stiffnesses and element lengths are obtained by optimization with a genetic algorithm. In addition to the standard optimization case, the effects of prestressed springs with contact release, nonlinear springs, optimizable initial configuration and an extra segment on the approximations are studied as well. Moreover, an extra objective function that concerns the distribution of energy among the springs is introduced. Eventually, the results that are obtained by the proposed method are verified with an experimental setup that contains a prototype of the system. The experimental results show agreement with the model with 93.47% work reduction. The corresponding model reduces the required work with more than 99%, which is higher than found in the state of the art.

This report is written for the thesis of my master Mechanical Engineering that was conducted at the department Precision and Microsystems Engineering of the Delft University of Technology. The goal of the project was to design a compliant hipback support of an exoskeleton while researching how to achieve desired directional compliance on this application. This work starts with an introduction to exoskeletons and explaining their working principles. The problem is that the reduction of range of motion for wearing an exoskeleton reduces the adoption of exoskeletons in industry. The objective of this study was to solve this by creating a hipback support with compliant behaviour in torsion and lateral bending while remaining stiff in the supporting forward bending direction. For inspiration, the state of the art of current hipback supports and research projects was reviewed. Also, the characterization methods of compliant mechanisms were reviewed for inspiration, to come up with a suitable design method for developing a compliant structure for a hipback support. By tuning stiffness of the structure in distinct directions, a conventional synthesis method for compliant mechanisms could be applied, namely the Freedom and Constraint Topologies method. This was used to come up with a preliminary design. This design was extended to improve stiffness behaviour for coupled motions by researching how to change stiffness on demand. The only categories to change stiffness are shape, material, prestress and boundary condition. For each category, a promising highlevel design was worked out with improved stiffness behaviour for coupled motions. To generate an even more satisfying solution, optimization of the beam shape was applied to handle the defined rather complex kinetostatic task and to develop a lightweight structure. The objective function was to reach specified deflections where shaperelated parameters are optimized. These parameters were the spine shape of the beam, the crosssection and orientation of the crosssection. To generate a design, a selfdeveloped optimizer using beam modelling was extended and applied. The extensions were to optimize a doubleclamped beam that was loaded in the middle point and optimizing for angular deformations as a set objective function. This extended optimizer generated the final design which was a thin rectangularshaped beam with threedimensional curves. This design was not straightforward to fabricate into a prototype. Multiple manufacturing methods were evaluated and one final method was selected to fabricate a close approximation of the final design. This final manufacturing method was to produce the prototype by roll bending and manually applying a twist on a curved lasercut strip which was extracted from flattening of the spatial shape. This prototype was verified by simulations and experiments. Simulation tests were run on an original design of the back support and compared with the new optimized shape based on the requirements. The results, for similar loadings and material, are 27 times more lateral flexibility, 1.25 more torsional flexibility and a design that is four times lighter in comparison to the original design. The original relative stiffness of 1 to 1 for forward relative to lateral bending improved to 2 to 1. An important remark to note is that the beam model generated a thin section which was less suitable for 1D beam modelling of the implemented FEM. Creating an optimizer based on 2D shell modelling code might give more accurate results and can be considered for future work.

Micrometer thin, braided wires, is a niche requirement that has applications in beam instrumentation. Due to the rare use-case of extremely thin braided wires there are no established or industrial braiding methods developed. For this reason, a novel method for braiding thin wires was developed. In order to achieve a repeatable result a prototype braiding machine was designed, built and tested. Stainless steel, nylon and carbon wires were tested and evaluated. The aim of this research was to prove that braiding extremely small yarns is possible. Also, the proposed machine parameters have been opti- mized to ensure repeatability and good quality wires production. By proving that braiding thin wires, in the order of a few tens of microns, we open the way for braiding extremely thin wires for wire scanners.

The objective of this thesis is to show that warping can be used in a thin-walled beam to create a suitable replacement for a classical differential mechanism. This normally negated deformation of the cross section can be used to create an inverse transmission mechanism, used to create a monolithic and simple alternative to a classic differential mechanism. This warping beam differential mechanism can be used to add walking functionality to a back support passive exoskeleton. These exoskeletons are optimized to reduce muscle stress during lifting, but this comes with the downside of an increase in muscle activity during walking. To counter this increase in muscle activity, the use of a differential mechanism is suggested. To show the feasibility of the proposed solution, a characterization of the warping beam differential mechanism is made for different geometric properties. This characterization is done with respect to two functionalities, to determine the geometric advantage of the inverse transmission and the rotational compliance under torque of the beam. The warping constant and torsion constant are determined to be the geometric properties of most influence on the behaviour of the warping beam. It is also shown that within the proposed boundary conditions, different beam cross sections with the same geometric properties show the same behaviour. An increase in warping constant results in an increase in geometric advantage and a reduction of the rotational compliance. The torsion constant has a linear correlation to the stored energy for the inverse transmission functionality, having less influence on the geometric advantage and rotational compliance compared to the warping constant. This thesis shows that this new kind of differential mechanism can be used to replace the classical mechanism and the potential can be seen with an achieved theoretical geometric advantage of almost 1. A systematic overestimation of the geometric advantage and an underestimation of the compliance is made by the model with respect to the experiments, but with some improvements this theoretical goal is not far away.