Living things are capable of incredible dexterity, agility, robustness, and elegance in their interactions with their environment. For far longer than the field of robotics has even existed, humanity has aspired to capture even a fragment of that dynamic elegance -- not merely out of curiosity, but also to inform the design of the artificial systems we hope to construct ourselves.

The field of soft robotics hypothesizes that by engineering comparably compliant, dynamically complex systems, some of this elegance can be translated to robotics. Thanks to recent strides in materials and fabrication, electrostatic actuators (ESAs) have emerged as a promising core technology for the field. These electrically-driven robotic muscles are powerful, fast, and can be made numerous geometries using a wide range of materials. However, electrostatic actuators are not without their own challenges. They operate at thousands of volts, but low currents -- a niche where few technologies exist.

This work aims to address a fundamental problem for any electrostatic actuator-based soft robot, and for the future of our field: how to translate between the theoretical and computational tools used to study system dynamics and synthesize global control laws, and the high voltage signals needed to independently drive these robotic muscles -- not one, or ten, but potentially hundreds to thousands of them. An answer to this question will open the door to the construction of highly performant soft robots with complex, coordinated motion, and facilitate the development of novel control techniques for these dynamically complex systems, taking us one step closer to the elegance of living things.

This work begins with a discussion on the current state of soft robotics, and identifies the limitations on the field imposed by a lack of actuators and accompanying drivers which are both performant and highly parallelizeable.We identify the potential of electrostatic actuation (ESA) to be a technology platform capable of resolving these limitations contingent on the development of parallelizeable high voltage driver circuitry and feedback controls, and set our objective to do so.

In the second chapter, we build a dynamical model of an electrostatic actuator from first principles and geometry.We explore the physics and dynamics of these actuators along with the assumptions which can be used to simplify their modelling. The result is a linearized model of an ESA driven by a bridge-style switching amplifier which is tractable for control design.

In the third chapter, we introduce a parallelizeable scheme for feedback control of electrostatic actuators. We introduce a parallelizeable optoelectronic driver circuit suitable for driving high voltage capacitive loads such as ESAs. We present a procedure to refine our derived dynamical model through the system identification of physical actuators. Using this information, we design a control law for feedback regulation of actuator voltage, and demonstrate and validate its implementation on a benchtop system.

The fourth chapter, we demonstrate the utility of our approach by achieving parallel, independent control of soft actuators at unprecedented scales. We present the construction of a synthetic tissue -- a soft, intelligent, robotic material with 100 independent channels of tightly integrated sensing, actuation, and control. Our system includes distributed magnetometers to leverage our insights from modelling ESA dynamics to implement both feedback control of actuator displacement and the sensing of externally applied forces. This chapter concludes with a discussion of the complex emergent behaviors that emerge as a result.

In the final chapter, we consider how to scale our approach by additional orders of magnitude. To do so, we present the design of a novel solid-state high-voltage switch based on series stacks of low-voltage transistors. Our design is more efficient, more performant, and better suited for mass manufacturing than existing alternatives. We introduce the theory of it's operation, discuss it's implementation, evaluate it's performance against the state of the art, while also laying a framework for future studies.