Tensegrity is a structural form that is defined as a set of rigid elements suspended in a net of continuous tension. This structure shows potential for compliance, impact tolerance and mechanical robustness. However, its non-linear coupled dynamics and often complex geometry require advanced control strategies. An actuator reference planning strategy to bring tensegrity robots closer to controlled full body movements was proposed by Guido Tournois \cite{GuidoThesis} in 2017. This strategy, called the Full Body Reference Planner (FBRP), finds a sequence of equilibrium configurations for a tensegrity structure, predominantly to follow a trajectory in space. However, the method is incapable of incorporating inequality constraints while obtaining said equilibrium configurations. This is a problem when dealing with certain restrictions, e.g., actuator limitations and stability of the structure.

In this thesis we implemented a robust way to account for inequality constraints while utilizing the FBRP. That was done by means of optimization, i.e., an implementation of a Sequential Quadratic Programming method to ensure inequality constraints were respected for each configuration. The approach was validated in scenarios related to practical applications where inequality constraints were enforced. The results showed advancements towards practical feasibility. Furthermore, the robustness, efficiency and accuracy of the method were validated. The extended implementation depicted robustness to parameter variations and good results in terms of accuracy. However, given the iterative nature of the method, it was more computationally expensive than its precursor.

Twisted and Coiled Polymer Actuators (TCPAs) are a new type of thermally driven fiber artificial muscles with many attractive advantages, including high power-to-weight ratio, large contractions, lightweight, and low cost. However, one drawback is a low force that one TCPA can generate. To fully implement TCPAs in robotic systems, TCPAs must be combined into larger structures to achieve higher forces. A common force amplification strategy, arranging actuators in parallel, can lead to bulky structures, especially for TCPAs. This design strategy differs from that in nature. Nature arranges muscle fibers in angular orientation, which is a space-saving strategy, to build powerful muscles in a slender muscle-like form factor. Subsequently, control of TCPA structures requires a model. Closely packing multiple TCPAs potentially introduces a thermal interaction between the TCPA fibers in a structure that could affect their force production.
This work investigates whether the behavior of a bioinspired TCPA structure can be predicted given a force model of single TCPA fiber. To that end, this work studies the influence of the neighboring thermal effects between TCPAs. The Joule-heated TCPAs were modeled using a Standard Linear Solid model in combination with a proportional contribution by temperature. Experiments were conducted to determine the influence of neighboring effects, by varying (1) the distance between fibers and (2) the number of fibers. Planar structures composed of up to 8 TCPAs were experimentally investigated. The scaled and geometrically-translated Joule-heated TCPA force model was used to compare the mechanical behavior of structures with different conditions.
The tested structures showed no notable difference in force measurements. It was concluded that the varied number of fibers and the distance between fibers in the TCPA structures did not affect force production of individual TCPA fibers, meaning that neighboring effects can be neglected. When heated, however, the thermodynamics of the structures could not be predicted by the TCPA force model, which was estimated in a separate experiment. Adaptations to convection coefficient in the TCPA force model resulted in sufficient prediction of the TCPA structures. The difference in model convection coefficient between experiments highlighted the TCPAs' sensitivity to conditions in the working environment. It was suggested that the TCPA structures up to 8 fibers could be adequately predicted given TCPA force model estimated under the exactly same ambient conditions as used in the structure.
This work provides the basis for TCPA structure modeling to promote the use of TCPAs in the wider force range robotic applications.

A humanoid robot is being developed, intended to operate robustly in a typical day-to-day human environment. Its ability to react quickly to stochastic disturbances, especially its ability to avoid falling, termed stability in this context, is of prime importance. The choosing of a suitable actuator that affords such stability is seen as a challenge in the current stage of the humanoid project. A novel method for comparing various actuators in terms of humanoid stability is investigated, to aid with this choice.
A common method for comparing actuators is the use of low-level actuator parameters, e.g. power density. With a low-level parameter, a specific actuator property can be compared between multiple actuators. To incorporate relevant properties of all actuators, a large set of actuator parameters is required. This becomes even more challenging when comparing a heterogeneous collection of actuators, e.g. muscle and hydraulics. Besides subjectiveness in parameter definitions and comparisons, the multidimensional set of actuator parameters needs to be related to a one-dimensional measure of humanoid stability. As low-level parameters and high-level humanoid stability are not directly relatable, subjective weighting factors are traditionally employed to reduce dimensionality.
Step time is introduced as a one-dimensional criterion for comparing different actuators. The ability to make fast movements is a critical component in defining humanoid stability. Therefore, by comparing step times of different actuators in a fixed test case, a measure of humanoid stability is obtained. The step movement represents a simplified motion of the lifting of a foot, in which essential elements of humanoid motion are incorporated.
A framework is developed, with which step times of Human Muscle (HM), ElectroMechanical Actuators (EMAs) and Hydraulic Actuators (HAs) are computed. While EMAs are often thought to have limited potential for humanoid applications, step times are similar to those of muscle. Supplemented with recent examples of successful EMA implementations, step time demonstrates the potential of EMAs. Furthermore, HA capability is also demonstrated, as step times can obtain lower values than those of HM and EMA.
Step time as developed in this thesis appears to be a powerful tool for comparing vastly different actuators. It provides a one-dimensional measure related to humanoid stability, evaluated directly for each specific actuator.

The Twisted and Coiled Polymer Muscle (TCPM) is a new, light weight compliant actuator that is easy and inexpensive to produce. As a result there is a growing body of research, with the goal of implementation of the TCPM. Our research combines two existing fields: 1. Self-sensing, eliminating the need for an external force and displacement sensor. 2. Structures of multiple TCPMs. This research compares the effect of a series versus parallel electrical configuration on self-sensing in Joule heated TCPMs. Experiments show that both series and parallel connection are fit for self-sensing. Relative errors in series configuration are 6.7% 6.9% and 5.1% for respectively strain, temperature and force. Relative errors in parallel configuration are 8.3% 10.4% and 7.8%. These data show series configuration performs better on self-sensing accuracy. This is due to the 2-3dB difference in signal to noise ratio, in favour of series. In addition good repeatability was found in the mechanical behaviour of the TCPMs, with a variance in the spring constant that is smaller than 10% for all cases. This research therefore demonstrates good repeatability as well as accurate self-sensing in a structure in favour of a series electrical configuration.