Doctor of Philosophy (Ph.D.), University of Dayton, 2023, Electrical Engineering

Autonomous systems usually require accurate localization methods for them to navigate safely in indoor environments. Most localization methods are expensive and difficult to set up. In this work, we built a low-cost and portable indoor location tracking system by using Raspberry Pi 4 computer, ultra-wideband (UWB) sensors, and inertial measurement unit(s) (IMU). We also developed the data logging software and the 2D Kalman filter (KF) sensor fusion algorithm to process the data from a low-power UWB transceiver (Decawave, model DWM1001) module and IMU device (Bosch, model BNO055). Autonomous systems move with different velocities and accelerations, which requires their localization performance to be evaluated under diverse motion conditions. We built a dynamic testing platform to generate not only the ground truth trajectory but also the ground truth acceleration and velocity. In this way, our tracking system's localization performance can be evaluated under dynamic testing conditions. The novel contributions in this work are a low-cost, low-power, tracking system hardware and software design, and a 2D linear stage experimental setup to observe the tracking system's localization performance under different dynamic testing conditions. The 2D testing platform has a 1 m translation length and 80 micrometers of bidirectional repeatability. The tracking system's localization performance is evaluated under dynamic conditions with eight different combinations of acceleration and velocity. The ground truth accelerations varied from 0.6 to 1.6 m/s$^2$ and the ground truth velocities varied from 0.6 to 0.8 m/s. Our experimental results show that the location error can reach up to 50 cm under dynamic testing conditions when only relying on the UWB sensor, with the KF sensor fusion of UWB and IMU, the location error decreases to 13.7 cm. For autonomous systems which require 3D real-time locating service, a 3D tracking device is designed based on the previously mentioned 2D track (open full item for complete abstract)

Committee: Vamsy. Chodavarapu Prof. (Advisor); Manish. Kumar Prof. (Committee Member); Raul. Ordonez Prof. (Committee Member); Vijayan K. Asari Prof. (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, The Ohio State University, 2012, Mechanical Engineering

This research involved the modeling, control and state estimation of a Roll Simulator. The focus of this study was on the Roll Simulator's application in emulating rollovers for vehicles such as ROVs. The Roll Simulator was designed to study occupant kinematics during a vehicle rollover in a laboratory setting. Little research has been performed where the focus has been on the vehicle rolling over to 90 degrees and the interaction of the occupant with the road plane at this instance has been closely examined. The Roll Simulator allows for these types of analyses to occur. In this dissertation, a two (2) degree-of-freedom model, describing the dynamics of the Roll Simulator, is developed. Equations of motion, derived using Lagrange's energy methods, describe the dynamics of the sled-platform assembly. Additional sub-system modeling is also performed to capture the dynamics of a hydraulic system, electro-magnetic particle brake and electric roll motor. The validity of the full simulation is corroborated by comparisons with experimental data from the Roll Simulator. Control strategies for the Roll Simulator are also discussed. The strategies are derived utilizing simple physics of the system. This allows for desired trajectories to be met using feed-forward terms. Application of feedback is limited due to the configurations of the actuators and the short duration maneuever. A variety of linear observers are introduced to estimate states within the Roll Simulator. A Kalman Filter is developed to estimate sled speed. To tune the filter, the Kalman Filter is applied to a higher fidelity model which has four (4) degrees-of-freedom. To capture the non-linear behavior of the sled-platform assembly, an Extended Kalman Filter (EKF) is used. When applied to experimental data, the observed sled speed exhibits gross over-estimation of the true speed. This is due to a disturbance in the system. A disturbance observer is used to estimate rolling resistance between the sled and f (open full item for complete abstract)

Committee: Dennis Guenther (Advisor); Gary Heydinger (Committee Member); Ahmet Kahraman (Committee Member); Gary Kinzel (Committee Member) Subjects: Mechanical Engineering

Master of Science in Electrical Engineering, Cleveland State University, 2010, Fenn College of Engineering

Surface micromachined low-capacitance MEMS capacitive accelerometers which integrated CMOS readout circuit generally have a noise above 0.02g. Force-to-rebalance feedback control that is commonly used in MEMS accelerometers can improve the performances of accelerometers such as increasing their stability, bandwidth and dynamic range. However, the controller also increases the noise floor. There are two major sources of the noise in MEMS accelerometer. They are electronic noise from the CMOS readout circuit and thermal-mechanical Brownian noise caused by damping. Kalman filter is an effective solution to the problem of reducing the effects of the noises through estimating and canceling the states contaminated by noise. The design and implementation of a Kalman filter for a MEMS capacitive accelerometer is presented in the thesis in order to filter out the noise mentioned above while keeping its good performance under feedback control. The dynamic modeling of the MEMS accelerometer system and the controller design based on the model are elaborated in the thesis. Simulation results show the Kalman filter gives an excellent noise reduction, increases the dynamic range of the accelerometer, and reduces the displacement of the mass under a closed-loop structure.

Committee: Lili Dong (Committee Chair); Charles Alexander (Committee Member); Siu-Tung Yau (Committee Member) Subjects:

Master of Sciences (Engineering), Case Western Reserve University, 2013, EECS - Electrical Engineering

Complete-coverage robots, such as a lawnmower or snowplow, require a centimeter-level localization solution in order to navigate reliably. Unmodeled wheel slip or other odometry errors may cause localization to diverge beyond the bounds of uncertainty. Specifically in the case of a robot snowplow, errors due to wheel slip may be significant. This thesis uses the CWRU Cutter autonomous robot as a test platform to address the dual issues of (1) robust localization and (2) odometry error handling. Both an Extended Kalman Filter and an Adaptive Monte Carlo Localization procedure are derived and implemented specifically on the CWRU Cutter robot. Finally, a new augmented Extended Kalman Filter with general-purpose wheel-velocity error states is derived. The augmented EKF is shown to fully estimate the robot state and the wheel velocity error due to wheel slip during logged data from the 2013 Institute of Navigation's Autonomous Snowplow Competition.

Committee: Roger Quinn Dr. (Advisor); Francis Merat Dr. (Committee Member); M. Cenk Cavusoglu Dr. (Committee Member) Subjects: Computer Engineering; Electrical Engineering; Robotics

Master of Science in Computer Engineering, University of Dayton, 2024, Electrical and Computer Engineering

Administering the behavior of drivers near roadside restricted areas, such as work zones, accident zones, or natural calamity zones, is necessary for safety. It helps steer vehicles clear of the ongoing blocked region. This ensures the safety of both drivers and people in that area. The vehicles need to be diverted to a different lane away from the restricted area for smooth running of the traffic. A computer vision-based autonomous system could be able to automatically monitor the movements of the vehicles and predict their pathways based on the direction and speed of the vehicles. This would help to provide appropriate signals to the drivers for changing the lanes appropriately. Development of an artificial intelligence-based learning system for detection and tracking vehicles on the road and prediction of their future locations in real-time videos captured by a stationary camera is proposed in this thesis. The videos captured in outdoor environments will be subjected to several challenges due to varying lighting conditions and changes in orientation, viewing angle, and object size. Surrounding objects like trees, buildings, or other vehicles can obscure a vehicle completely or partially, making reliable detection and tracking difficult. Stationary cameras may also capture background regions like buildings, trees, parking lots, etc. Sometimes, the detection vehicles become difficult due to their darker texture in non-uniform lighting conditions. In this thesis research, a YOLO_v8 neural network model is employed to detect the vehicles in the video frames in real-time. The neural network model needs an extensive set of annotated datasets of vehicles in roadside environments. A new annotated dataset named Dayton Annotated Vehicle Image Set (DAVIS) suited for US road conditions is built to train the vehicle detection model. An adaptive image enhancement technique, namely Contrast Limited Adaptive Histogram Equalization (CLAHE) is used in the moving object regions (open full item for complete abstract)

Committee: Vijayan K. Asari (Committee Chair); Theus Aspiras (Committee Member); Eric J. Balster (Committee Member) Subjects: Computer Engineering

MS, University of Cincinnati, 2024, Engineering and Applied Science: Aerospace Engineering

Autonomous vehicles are a hot topic in control theory and are utilised in different fields such as industries, aerospace and robotics. Trajectory-tracking is one of the crucial features of autonomous vehicles which involves two major steps: generating a reference trajectory and tracking it. In many cases, the reference trajectory is generated by using an objective function where we want vehicles to move towards the extremum (maximum/minimum) of the objective function which is a terminal position of the trajectory. We performed some simulations and experiments using traditional controllers including the Proportional-Derivative Controller (PDC), Model-Predictive Controller (MPC), and Pure Pursuit Controller (PPC). These controllers while showing some degree of desirable vs. undesirable behaviour, they are model-based controllers that require a mathematical expression of an objective function. Additionally, they show difficulties and they have other limitations. In this thesis, we make a case for the utilization of extremum-seeking control (ESC) systems which are model-free, real-time, adaptive systems. We revisit some works that have been done regarding the classic ESC (C-ESC) structure. The primary part of this thesis is where we provide the simulations and experimental works using control-affine ESC (CA-ESC) systems which have been used rarely in experimental environments in literature. Particularly, we utilized single-integrator and unicycle dynamics CA-ESC structures and conducted simulations and experiments. Additionally, we propose a novel amended CA-ESC design (for both single-integrator and unicycle dynamics) by adopting some developments that took place in this area in recent years. Our proposed design consists of a Geometric-Based Extended Kalman Filter (GEKF) for gradient estimation and adaptation law for attenuation oscillations and better convergence rate. We performed simulations to show the effectiveness of our proposed design.

Committee: Sameh Eisa Ph.D. (Committee Chair); Shaaban Abdallah Ph.D. (Committee Member); Abhinav Sinha Ph.D. (Committee Member) Subjects: Aerospace Materials

Master of Science, The Ohio State University, 2024, Biostatistics

Accurate forecasting of weekly number of influenza (flu) lab tests and positive cases is vital for hospitals to provide adequate patient care at the right time. It also helps prevent shortages or overages of staffs and supplies. In this paper we present a practical implementation of a Bayesian Kalman filter to forecast weekly flu test and positive cases in a hospital environment. By integrating real time hospital data, this framework offers a robust methodology for accurately predicting flu volume one to four weeks out with a reasonable accuracy.

Committee: Grzegorz Rempala (Advisor); Eben Kenah (Committee Member); Fernanda Schumacher (Committee Member) Subjects: Biostatistics; Health Care Management; Mathematics; Medicine; Statistics

Master of Science in Electrical Engineering (MSEE), Wright State University, 2023, Electrical Engineering

Simultaneous Localization and Mapping (SLAM) is the process of using distance measurements to points in the surrounding environment to build a digital map and perform localization. It has been observed that featureless environments like tunnels or straight hallways will cause positioning faults in SLAM. This research investigates the fault diagnosis and accommodation problem for a laser-rangefinder-based SLAM systems on a quadrotor. A potential solution of using optical flow as velocity estimate and an extended Kalman filter (EKF) to perform position estimation is proposed. A fault diagnosis method for detecting faults in positional SLAM data or optical flow velocity data is developed by using two parallel EKFs. When a fault in the SLAM position or optical flow velocity is detected, the EKF adapts to provide a robust position estimate to ensure the safety of the flight control system.

Committee: Xiaodong Zhang Ph.D. (Advisor); Zhiqiang Wu Ph.D. (Committee Member); Weisong Wang Ph.D. (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, The Ohio State University, 2023, Mechanical Engineering

Metal Powder Bed Fusion (PBF) is a type of additive manufacturing process that incrementally builds parts by fusing 2D slices of the geometry into layers of metal powder, using either a laser (L-PBF) or electron beam (E-PBF), and is among the emerging technologies of Industry 4.0. The predominant quality control methods for PBF are pre- and post-process tests of the part and materials, which are inefficient because they cannot prematurely halt malfunctioning builds as errors occur. Live (In-situ) monitoring of the PBF process for defects, in which the defects are oftentimes due to improper thermal management, and in-situ control of the PBF process to ensure good thermal management, are areas of active research. These efforts are currently dominated by constructing data-driven PBF thermal models and using the corresponding estimations to judge the current thermal state (process monitoring) and to decide correction factors (process control). Collecting the data for training these methods is costly and renders them inflexible with respect to changes in part design and processing conditions, because they do not offer guaranteed performance in environments that lay outside the scope of the training data. Since PBF exists to increase production flexibility, lessening this dependency on training data is essential. To address this challenge, we demonstrate the efficacy of applying training data-free algorithms to the in-situ PBF thermal process monitoring and control problems. Our process monitoring algorithm is the Ensemble Kalman Filter (EnKF), which is a type of state estimator that uses a particle swarm to generate self-tuned, approximately 2-norm optimal, model-based estimates of the relevant process signatures. Here, the signatures are all temperatures in the PBF build. Our control algorithm is Model Predictive Control (MPC), which uses model-based predictions of future process signatures (here, temperatures) to determine a sequence of process inputs that reg (open full item for complete abstract)

Committee: David Hoelzle (Advisor); Andrew Gillman (Committee Member); Andrea Serrani (Committee Member); Mrinal Kumar (Committee Member); Michael Groeber (Committee Member) Subjects: Mechanical Engineering

Master of Science, University of Toledo, 2022, Electrical Engineering

The push for renewable energy has certainly driven the world towards sustainability. However, the incorporation of clean energy into the electric power grid does not come without challenges. When synchronous generators are replaced by inverter based Photovoltaic (PV) generators, the voltage profile of the grid gets considerably degraded. The effect in voltage profile, added with the unpredictable generation capacity, and lack of good reactive power control eases opportunities for sneaky False Data Injection (FDI) attacks that could go undetected. The challenge is to differentiate these two phenomena. In this thesis work, an attack is exposed in a grid environment with high PV penetration, and challenges associated with designing a detector that accounts for inefficiencies that comes with it is discussed. The detector is a popular Kalman Filter based anomaly detection engine that tracks deviation from the predicted behavior of the system. Chi-squared fitness test is used to check if the current states are within the normal bounds of operation. The work concludes by exposing a vulnerability in using static and dynamic threshold detectors which are directly affected by day-ahead demand prediction algorithms that have not been fully evolved yet. Finally, some of the widely used machine learning based anomaly detection algorithms is used to overcome the drawbacks of model-based algorithm.

Committee: Weiqing Sun (Committee Chair); Ahmad Javaid (Committee Member); Junghwan Kim (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, Case Western Reserve University, 2021, Applied Mathematics

The goal of the electrical impedance tomography (EIT) inverse problem is to estimate the electrical conductivity distribution inside a body from the body's response to applied voltages or injected electric currents along its boundary. Mathematically, the response can be described in terms of the Dirichlet-to-Neumann (DtN) operator of the governing conductivity equation. While several direct reconstruction methods assume that the DtN map is available, in practice, the EIT measurement data collected by using a finite number of contact electrodes provide only the knowledge of the resistance matrix, a mapping between the applied current patterns and the corresponding measured voltages at the electrodes, referred to as the electrode data. The DtN data and the electrode data correspond to different boundary value problems, and the direct connection between them is not straightforward. The first contribution of this thesis is to explore the relation between the two boundary data and propose a method to approximate DtN data from the resistance matrix that can be computed from the electrode data. The problem is ill-posed and its numerical treatment requires regularization. The proposed method, formulated as an inverse problem in the Bayesian framework, leverages a sample-based prior and a principal component model reduction. In the second contribution of this thesis, we explore a direct reconstruction method for solving the EIT inverse problem, known as the layer stripping method, where the data consist of the Neumann-to-Dirichlet operator on the boundary. We recast the layer stripping process in the Bayesian framework, reformulating it as a state estimation problem, and propose an algorithm for its numerical solution based on Ensemble Kalman Filtering (EnKF). Our novel Bayesian layer stripping method is quite robust, derivative-free, and intrinsically suited for the quantification of uncertainties in the estimate. Moreover, the results of reconnecting the discrete (open full item for complete abstract)

Committee: Erkki Somersalo Prof (Advisor); Daniela Calvetti Prof (Advisor); Steven Izen Prof (Committee Member); Michael Martens Prof (Committee Member) Subjects: Applied Mathematics

MS, University of Cincinnati, 2021, Engineering and Applied Science: Mechanical Engineering

The aim of this thesis is to present the development of two path predicting algorithms for small Unmanned Aerial Systems (sUAS) operating within the national airspace. As sUAS have grown in popularity, the risk of mid-air collisions has increased. The number of these vehicles in use will continue to grow exponentially as restrictions imposed by the Federal Aviation Administration are removed. Many of these vehicles will be performing fully autonomous and beyond visual line of sight missions. This thesis covers the development and testing of two real-time path predicting algorithms which generate future trajectories which can be used to identify when a collision is likely. The system presented in this thesis can be used to alert operators or the vehicles themselves to the pending collision to make corrective actions. The path predictors developed allow for collisions to be detected sooner. The increase in time before collision will ensure that a safe minimum separation distance between vehicles is maintained. The additional time also allows for more optimal rerouting measures to be taken. Predictions have been generated up to thirty seconds into the future, however, have higher uncertainty when compared to a fifteen or twenty second position estimate. The path prediction algorithms make use of real time position data broadcast by the vehicles while in flight. The vehicles will be required to broadcast their position in terms of latitude, longitude, altitude, unique identification number and their airspeed. The system discussed here uses the data being broadcast and not the waypoints in the mission being flown. The system developed for this thesis will not require any special modifications to the vehicles. This system is run on a ground control station and does not require on-board computation by vehicles. The first path prediction algorithm developed and discussed in this thesis is based on a multiple model tracker. This uses a constant velocity and con (open full item for complete abstract)

Committee: Manish Kumar Ph.D. (Committee Chair); Zachariah Fuchs Ph.D. (Committee Member); Rajnikant Sharma Ph.D. (Committee Member); David Thompson Ph.D. (Committee Member) Subjects: Mechanical Engineering

MS, University of Cincinnati, 2020, Engineering and Applied Science: Electrical Engineering

Taking inspiration from the success of previously established deep learning based tracking frameworks (such as Recurrent Convolutional Neural Network detectors), we propose a novel fine-tuned object detection and tracking network which works based on a CNN feature-based deep tracker approach along with the widely popular Faster Region-based Convolutional Neural Network feature detector. We implement a fine-tuned 2 stage Multi Object Detection and Tracking network integrated together. More specifically, the first part of the network is an improved Faster Region-based Convolutional Neural Network based object detection module that has been trained on our aerial vehicle image datasets. The second part of the network is an improved CNN feature-based deep tracker module that is a combination of a multi object tracker and a feature detection CNN that can be used for re-identification of vehicles. This type of network drastically reduces the computational need of training the tracker with pre-trained CNNs and significantly improve the accuracy of tracking in aerial traffic videos by using the unique re-identification feature and priority match algorithm. The proposed method has been trained and then tested using aerial videos obtained from UAVs operated by the UAV MASTER lab at University of Cincinnati. These videos represent both interrupted (obtained at intersections) and uninterrupted (obtained at highways) traffic conditions.

Committee: Manish Kumar Ph.D. (Committee Chair); Anca Ralescu Ph.D. (Committee Member); Ranganadha Vemuri Ph.D. (Committee Member) Subjects: Artificial Intelligence

Doctor of Philosophy, The Ohio State University, 2020, Electrical and Computer Engineering

Inertial sensors became wearable with the advances in sensing and computing technologies in the last two decades. Captured motion data can be used to build a pedestrian inertial navigation system (INS); however, time-variant bias and noise characteristics of low-cost sensors cause severe errors in positioning. To overcome the quickly growing errors of so-called dead-reckoning (DR) solution, this research adopts a pedestrian INS based on a Kalman Filter (KF) with zero-velocity update (ZUPT) aid. Despite accurate traveled distance estimates, obtained trajectories diverge from actual paths because of the heading estimation errors. In the absence of external corrections (e.g., GPS, UWB), map information is commonly employed to eliminate position drift; therefore, INS solution is fed into a higher level map-matching filter for further corrections. Unlike common Particle Filter (PF) map-matching, map constraints are implicitly modeled by generating rasterized maps that function as a constant spatial prior in the designed filter, which makes the Bayesian estimation cycle non-recursive. Eventually, proposed map-matching algorithm does not require computationally expensive Monte Carlo simulation and wall crossing check steps of PF. Second major usage of the rasterized maps is to provide probabilities for a self-initialization method referred to as the Multiple Hypothesis Testing (MHT). Extracted scores update hypothesis probabilities in a dynamic manner and the hypothesis with the maximum probability gives the correct initial position and heading. Realistic pedestrian walks include room visits where map-matching is de-activated (as rasterized maps do not model the rooms) and consequently excessive positioning drifts occur. Another MHT approach exploiting the introduced maps further is designed to re-activate the map filter at strides that the pedestrian returns the hallways after room traversals. Subsequently, trajectories left behind inside the rooms are heuristically adjus (open full item for complete abstract)

Committee: Alper Yilmaz Prof (Advisor); Keith Redmill Prof (Committee Member); Charles Toth Prof (Committee Member); Janet Best Prof (Other) Subjects: Electrical Engineering; Engineering

Master of Arts (MA), Bowling Green State University, 2020, Mathematics

Control and estimation theory are branches of mathematics that involve using data and measurements to estimate the value of a parameter of interest, and how changing certain parameters effects this estimation. The Kalman filter is a fundamental result in control and estimation theory that was introduced by Rudolf E. Kalman in 1960. The Kalman filter is a set of equations that provides an optimal estimate of the state of a system in a least-squares sense. The filter is often sought for its recursive and noise-smoothing properties, and has been found useful across many disciplines and in real world systems. This thesis will contribute to the literature of control and estimation theory by providing an introduction to the principles of the filter. This introduction includes a brief history of the filter, a derivation of the filter equations, and simple examples of applications of the filter.

Committee: Kit Chan Dr. (Advisor); So-Hsiang Chou Dr. (Committee Member) Subjects: Mathematics

MS, University of Cincinnati, 2019, Engineering and Applied Science: Mechanical Engineering

Every year, people in physically strenuous fields such as nursing are at risk of work-related injury due to unhealthy, unsafe, and repetitive movements, such as lifting and moving heaving objects. To reduce this risk, we present a novel approach to collecting movement data by applying an Unscented Kalman Filter to the data online data collection from seven inertial measurement units (IMU) located above and below the ankle, knee, and hip joints, allowing the estimation of joint angles. These sensors are incorporated into a developed pair of leggings that has channels to route the wiring through but allows for the sensors to be easily accessed via pockets, but firmly mounted onto the user. The data from the IMU's are read by an Arduino and is then transmitted to a Raspberry pi, which filters the incoming data. Once the recording is complete, the data is printed to csv files. Points along the user's body will be calculated using the Denavit-Hartenberg method.

Committee: Manish Kumar Ph.D. (Committee Chair); Tamara Lorenz Ph.D. (Committee Member); David Thompson Ph.D. (Committee Member) Subjects: Engineering

MS, University of Cincinnati, 2019, Engineering and Applied Science: Aerospace Engineering

Autonomous vehicles have become a reality in many military and civilian applications. The ability to deploy them in constrained environments, such as regions with limited or no Global Positioning System (GPS) access, or no a-priori map information, can not only further their application space but also add capability to successfully complete tasks that are otherwise considered dull, dirty or dangerous for humans. This Thesis proposes and implements an autonomous navigation solution for an Unmanned Aerial Vehicle (UAV) using a robot-centered reference frame. An Extended Kalman Filter (EKF) is used to estimate the relative orientation of the UAV with respect to an user-defined goal and other objects of interest in the environment (called landmarks). A visual tracker continuously tracks these objects and based on the camera parameters, calculates their bearing measurement with respect to the UAV. This method uses a bearing-only measurement model to update the states of the system. The goal is selected real-time from the video feed provided by the UAV's onboard camera and the UAV has to navigate to it while avoiding obstacles along its path. A combined PN-guidance and obstacle avoidance controller is used for this purpose. A detailed 2D observability analysis is performed to find the sufficient conditions required for the system to be observable. The problem formulation is corroborated through extensive simulation and hardware results.

Committee: Rajnikant Sharma Ph.D. (Committee Chair); Manish Kumar Ph.D. (Committee Member); Ou Ma Ph.D. (Committee Member) Subjects: Engineering

Master of Science in Electrical Engineering (MSEE), Wright State University, 2018, Electrical Engineering

In this thesis, we consider the problem of tracking in complex nonlinear dynamical systems. While the Kalman filter is known to be the mean-squared error optimal tracker under linear dynamics and linear measurements, more sophisticated models and algorithms are required for complex dynamics. Here, we consider switching systems where the dynamical properties vary (''switch modes") over time. For example, the dynamics of a vehicle may switch as it transitions from interstate to urban conditions, human speech dynamics switch as speakers change, and stock market dynamics switch with discrete news events. In this work, we use mode-dependent neural networks to capture different nonlinear dynamics in a given system, and we developed a new algorithm, dubbed the Switching Neural Network Tracker (SNNT), to track modes and states over time. The proposed Bayesian system includes Markovian dynamics to model mode transitions and employes the Unscented Transform to mitigate computational complexity while estimating posterior probabilities. Examples using synthetic robot motion data and measured honeybee dance data demonstrate accurate mode identification and a dramatic reduction in state estimation error relative to non-switching systems.

Committee: Joshua Ash Ph.D. (Advisor); Arnab Shaw Ph.D. (Committee Member); Fred Garber Ph.D. (Committee Member) Subjects: Electrical Engineering

MS, University of Cincinnati, 2018, Engineering and Applied Science: Mechanical Engineering

This thesis focuses on localization of mobile robots in indoor environments without the use of pre-deployed sensor networks. The localization of mobile robots in indoor environment is very difficult due to the absence of Global Positioning System (GPS) signals. The problem of localization in indoor environments is usually solved using Simultaneous Localization and Mapping (SLAM) algorithms. However, these algorithms often prove to be insufficient in complex and dynamic environments. An example of such environment is a tunnel which does not provide distinguishing environmental features for the SLAM algorithms to work properly. The absence of visible light makes it difficult to use visual sensors such as cameras. In such environments, without the use of pre-deployed sensor networks, it is very difficult to obtain localization of the robot. This thesis proposes the use of active deployment of landmarks by the robot itself. The robot is assumed to have a physical capacity of carrying Radio Frequency (RF) Beacons which are deployed in the environment based on the calculations of the predicted co-variance of position error. The robot tries to achieve its goal based on the combination of data from the encoder and RF beacons. The system of transmitting RF beacons is deployed by the mobile robot which carries the receiver beacon as it moves through the environment. Using a combination of Dead Reckoning and tri-lateration of position using the RF beacons in the framework of Extended Kalman filter, the robot localized in the environment. As the RF beacons are deployed by the mobile robot, their locations are approximated using Levenberg- Marquardt algorithm. The mobile robot monitors the estimate of its localization error which is then used to make decisions to deploy successive beacons. The operative structure of the mobile robot is provided in the thesis which could be used to achieve desired navigation.

Committee: Manish Kumar Ph.D. (Committee Chair); Rui Dai Ph.D. (Committee Member); David Thompson Ph.D. (Committee Member) Subjects: Robots

Master of Science (MS), Ohio University, 2018, Electrical Engineering & Computer Science (Engineering and Technology)

Small Unmanned Aircraft Systems (sUAS) are becoming very popular for solving a multitude of problems. As sUAS solutions are applied to more often, it is evident that multiple cooperative sUAS can be beneficial to certain tasks (surveillance, inspection, mapping). Unfortunately, operations involving multiple sUAS are inherently complex, requiring navigation solutions that are very accurate both in a relative and absolute sense for every member of the swarm. This thesis presents a method to use ultra-wideband (UWB) range radios to increase the relative position accuracy (and as a byproduct, absolute position accuracy) of the members of a swarm. A range radio system is also developed and analyzed, allowing simulations for testing this method. Finally, real flight data has been collected using multiple custom-built sUAS platforms and post-processed, allowing the filter to be analyzed using real world data.

Committee: Maarten Uijt de Haag (Advisor); Michael Braasch (Committee Member); Frank Van Graas (Committee Member); Geoffrey Dabelko (Committee Member) Subjects: Electrical Engineering