This thesis presents rules that characterize the root locus for polynomials that are nonlinear in the root-locus parameter k. Classical root locus applies to polynomials that are affine in k. In contrast, this thesis considers polynomials that are quadratic or cubic in k. In particular, we focus on constructing the root locus for linear feedback control systems, where the closed-loop denominator polynomial is quadratic or cubic in k. First, we present quadratic root-locus rules for a controller class that yields a closed-loop denominator polynomial that is quadratic in k. Next, we develop cubic root-locus rules for a controller class that yields a closed-loop denominator polynomial that is cubic in k. Finally, we extend the quadratic root-locus rules to accommodate a larger class of controllers. We also provide controller design examples to demonstrate the quadratic and cubic root locus. For example, we show that the triple integrator can be high-gain stabilized using a controller that yields a closed-loop denominator polynomial that is quadratic in k. Similarly, we show that the quadruple integrator can be high-gain stabilized using a controller that yields a closed-loop denominator polynomial that is cubic in k.

Power converter has played a very important role in modern electric power systems. The control of power converters is necessary to achieve high performance. In this study, a dc-dc buck converter is studied. The parameters of a notional proportional-integral controller are to be selected. Genetic algorithms (GAs), which have been widely used to solve multi-objective optimization problems, is used in order to locate appropriate controller design. The control metrics are specified as phase margin in frequency domain and voltage error in time-domain. GAs presented the optimal tradeoffs between these two objectives. Three candidate control designs are studied in simulation and experimentally. There is some agreement between the experimental results and the simulation results, but there are also some discrepancies due to model error. Overall, the use of multi-domain, multi-objective-optimization-based approach has proven feasible.

This paper describes an analog-to-digital signal converter which varies its precision as a function of input slew rate (maximum signal rate of change), in order to best follow the input in real time. It uses Flash and Successive Approximation (SAR) conversion techniques in sequence. As part of the design, the concept of "total real-time optimization" is explored, where any delay at all is treated as an error (Error = Delay * Signal Slew Rate). This error metric is proposed for use in digital control systems. The ADC uses a 4-bit Flash converter in tandem with SAR logic that has variable precision (0 to 11 bits). This allows the Tandem ADC to switch from a fast, imprecise converter to a slow, precise converter. The level of precision is determined by the input’s peak rate of change, optimized for minimum real-time error; a secondary goal is to react quickly to input transient spikes. The implementation of the Tandem ADC is described, along with various issues which arise when designing such a converter and how they may be dealt with. These include Flash ADC inaccuracies, rounding issues, and system timing and synchronization. Most of the design is described down to the level of logic gates and related building blocks (e.g. latches and flip-flops), and various logic optimizations are used in the design to reduce calculation delays. The design also avoids active analog circuitry whenever possible – it can be almost entirely implemented with CMOS logic and passive analog components.

Autonomous harvesting may provide a viable solution to mounting labor pressures in the United States' strawberry industry. However, due to bottlenecks in machine perception and economic viability, a profitable and commercially adopted strawberry harvesting system remains elusive. In this research, we explore the feasibility of using deep reinforcement learning to overcome these bottlenecks and develop a practical algorithm to address the sub-objective of viewpoint optimization, or the development of a control policy to direct a camera to favorable vantage points for autonomous harvesting. We evaluate the algorithm's performance in a custom, open-source simulated environment and observe affirmative results. Our trained agent yields 8.7 times higher returns than random actions and 8.8 percent faster exploration than our best baseline policy, which uses visual servoing. Visual investigation shows the agent is able to fixate on favorable viewpoints, despite having no explicit means to propagate information through time. Overall, we conclude that deep reinforcement learning is a promising area of research to advance the state of the art in autonomous strawberry harvesting.

The state-of-art model-free reinforcement learning algorithms can generate admissible controls for complicated systems with no prior knowledge of the system dynamics, so long as sufficient (oftentimes millions) of samples are available from the environ- ment. On the other hand, model-based reinforcement learning approaches seek to leverage known optimal or robust control to reinforcement learning tasks by mod- elling the system dynamics and applying well established control algorithms to the system model. Sliding-mode controllers are robust to system disturbance and modelling errors, and have been widely used for high-order nonlinear system control. This thesis studies the application of sliding mode control to model-based reinforcement learning. Computer simulation results demonstrate that sliding-mode control is viable in the setting of reinforcement learning. While the system performance may suffer from problems such as deviations in state estimation, limitations in the capacity of the system model to express the system dynamics, and the need for many samples to converge, this approach still performs comparably to conventional model-free reinforcement learning methods.

Today’s digital communication systems perform data transfers at the rate of millions of bits per minute, with data errors in the order of l/6th error per day. This magnitude of errorless communication is now possible because of sophisticated error correcting codes. Many types of error codes are employed today in three distinct areas of digital data communication: human to computer; data source to computer; computer to computer; and intra-computer; we are concerned here with intra-computer communication. This research is primarily a mathematical study of error codes in general to explore the possibilities of each major type for the purpose of implementation in real systems. The author was inspired toward this goal by several people and self-feelings. The first, was a definite affinity toward orderliness and the logical sequence of formal mathematics. Secondly, the thrusting of being assigned to a work project where computer maintenance and where all types of errors became important. And, finally an advisor who believes in “practical things”. The original portion of this endeavor is to be found in the conclusions drawn from each group of mathematical facts disclosed in the research. The particular bend of the author toward the cost/reliability/ efficiency of the system was not the intent of the theoretical mathematicians who did the majority of the work quoted herein. The author's contribution was to draw these ideas and works together and to form the conclusions based upon his experience and training as an Engineer. The primary conclusion is that multi-residue systematic codes appear to be the best choice for implementation of all around error correction and general hardware configurations. This conclusion is within the constraints that were laid down in the introduction of the research; 1) to not increase the cost of hardware, 2) to maintain or improve the system reliability, and 3) to maintain or increase the processing speed.

An increase in demand for electric power has forced utility transmission systems to continuously operate under stressed conditions, which are close to instability limits. Operating power systems under such conditions along with inadequate reactive power reserves initiates a sequence of voltage instability points and can ultimately lead to a system voltage collapse. Significant research have been focused on time-synchronized measurements of power systems which can be used to frequently determine the state of a power system and can lead to a more robust protection, control and operation performance. This thesis discusses the applicability of two voltage stability synchrophasor-based indices from literature to analyze the stability of a power system. Various load flow scenarios were conducted on the BPA 10-Bus system and the IEEE 39-Bus System using PowerWorld Simulator. The two indices were analyzed and compared against each other along with other well-known methods. Results show that their performances are coherent to each other regarding to voltage stability of the system; the indices can also predict voltage collapse as well as provide insight on other locations within the system that can contribute to instability.

In this paper, the architecture and implementation of an embedded controller for a steering based semi-autonomous collision avoidance system on a 1/10th scale model is presented. In addition, the development of a 2D hardware-in-the-loop simulator with vehicle dynamics based on the bicycle model is described. The semi-autonomous collision avoidance software is fully contained onboard a single-board computer running embedded GNU/Linux. To eliminate any wired tethers that limit the system’s abilities, the driver operates the vehicle at a user-control-station through a wireless Bluetooth interface. The user-control-station is outfitted with a game-controller that provides standard steering wheel and pedal controls along with a television monitor equipped with a wireless video receiver in order to provide a real-time driver’s perspective video feed. The hardware-in-the-loop simulator was developed in order to aid in the evaluation and further development of the semi-autonomous collision avoidance algorithms. In addition, a post analysis tool was created to numerically and visually inspect the controller’s responses. The ultimate goal of this project was to create a wireless 1/10th scale collision avoidance research platform to facilitate human studies surrounding driver assistance and active safety systems in automobiles. This thesis is a continuation of work done by numerous Cal Poly undergraduate and graduate students.

Telepresence describes technologies that allow users to remotely experience the sensation of being present at an event without being physically present. An avatar exists to represent the user whilst in a remote location and is tasked to collect stimuli from its immediate surroundings to be delivered to the user for consumption. With the advent of recent developments in Virtual Reality technology, viz., head-mounted displays (HMDs), new possibilities have been enabled in the field of Telepresence. The main focus of this thesis is to develop a solution for visual Telepresence, where an HMD is used to control the direction of a camera‟s viewpoint, such that the user‟s head is tracked by the avatar, while providing visual feedback to the user. The design and development of the device follows a mechatronic approach, where a real time operating system (RTOS) is used in conjunction with a microcontroller for mechanical actuator control. The first-generation prototype, HOG-1 (HMD-Operated Gimbal, rev. 1), developed for this thesis serves as a foundation for study; the implementation and analysis of the prototype contributes to the state of the art by providing a clearer glimpse of hardware and software requirements that are necessary to construct an improved model. Additionally, qualitative and quantitative measurements are developed in the process of this research.

This thesis document outlines the development of a multibody dynamics simulation of an actively stabilized multiple-input, multiple-output, coupled, balancing cube and the process of verifying the results by implementing the control algorithm in hardware. A non-linear simulation of the system was created in Simscape and used to develop a Linear Quadratic Gaussian control algorithm. To implement this algorithm in actual hardware, the system was first designed, manufactured, and assembled. The structure of the cube and the reaction wheels were milled from aluminum. DC brushless motors were installed into the mechanical system. In terms of electronics, a processor, orientation sensor, motor drivers, analog to digital converters, and a pulse width modulation board were assembled into the cube. Upon completion, the software to control the cube was developed using Simulink and run on a Raspberry Pi computer within the mechanism.