Master of Science in Computer Engineering (MSCE), Wright State University, 2017, Computer Engineering

The traditional education systems that have been used for several centuries have evolved very slowly and might be ineffective for addressing diverse learning styles and levels of preparation. This system is characterized by many students interacting with a single teacher, who is unable to address the individual needs of every student. Therefore, some students can become frustrated and fail to reach their educational potential. An Intelligent Tutoring System (ITS), which is a computer application used to provide students with one-to-one supplemental tutoring tailored to the student's learning style and pace, is of interest to educators for improving student learning. To evaluate the effectiveness of ITS, a systematic review of the recent literature was performed using a carefully crafted protocol designed to provide data to support a meta-study of the effectiveness of ITS. The research question guiding this study is: “Does an intelligent tutoring system improve students' learning abilities more than traditional learning?” A t-test, one-way ANOVA test, and KNIME program that does Latent Dirichlet allocation were performed. The results support the conclusion that ITS causes a significant improvement in learning over traditional instructional methods.

Committee: Mateen Rizki Ph.D. (Advisor); Michael Raymer Ph.D. (Committee Co-Chair); Jack Jean Ph.D. (Committee Member) Subjects: Computer Engineering; Computer Science

PhD, University of Cincinnati, 2023, Engineering and Applied Science: Aerospace Engineering

Conventional control approaches have been developed based on mathematical models of systems that contain multiple user-defined parameters, and it is time-consuming to determine such parameters. With advancements in computing power, artificial intelligence (AI) has been recently used to control autonomous systems. However, it is difficult for engineers to understand how the resulting output is obtained because most AI techniques are a black box without defining a mathematical model. On the other hand, a fuzzy inference system (FIS) is a preferable option because of its explainability. By adding learning capability to the FIS using a genetic algorithm (GA), the FIS can provide a near-optimal solution, which is known as a genetic fuzzy system (GFS). To exploit the advantages of the GFS, this work develops the FIS-based control approaches for diverse autonomous platforms, which include aerial, ground, and space platforms. For aerial platforms, this work develops a FIS-applied collision avoidance (CA) algorithm that can provide a near-optimal solution in terms of the travel distance of unmanned aerial vehicles (UAVs). After introducing a compact form of equations, which reduces the number of unknown parameters from 6 to 2, based on the enhanced potential field (EPF) approach, the proposed FIS models determine two unknowns, which are the magnitude of the avoidance maneuvers. The proposed models are trained to overcome the drawbacks of the artificial potential field (APF) while minimizing the travel distance of the UAVs, the trained FIS models are tested in a complex environment in the presence of multiple static and dynamic obstacles by increasing the number of UAVs in a given area. Numerical simulation results are presented for the training and testing results, including the comparison with the EPF. For ground platforms, this work proposes a decentralized multi-robot system (MRS) control approach to perform a collaborative object transportation with a near- (open full item for complete abstract)

Committee: Donghoon Kim Ph.D. (Committee Chair); Anoop Sathyan Ph.D. (Committee Member); Ou Ma Ph.D. (Committee Member); Kelly Cohen Ph.D. (Committee Member) Subjects: Aerospace Engineering

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

Automated Vehicles (AVs) must choose optimal driving actions under various traffic situations. However, designing their decision-making systems is challenging because it's impossible to anticipate all possible traffic situations. Previous studies proposed handcrafting decision-making rules for expected situations or for training an approximated model, using a limited amount of data to enable AVs to make proper decisions. However, these approaches could not recognize unexpected situations, so their decision-making was incomplete. Although the model-free Reinforcement-Learning (RL) approach is gaining attention as an alternative method these limitations, the RL controller's performance is highly dependent on the selection of the reward function and the simulation data. Given the challenge of hand-crafting a perfect set of decision-making rules or reward functions, a novel online evolving method, named evolving Finite State Machine (e-FSM), is proposed to develop an optimal Markov driving model from scratch. Although the Markov driving model has the same properties as the standard Markov model, its states and transition dynamics evolve by determining new states and identifying transition dynamics in real-time, without human supervision. To Implement the Markov model evolved by the e-FSM, two approaches have been studied for directly and indirectly supporting the RL controller's decision-making. First, an Online Evolving Framework (OEF), which consists of an RL controller, the e-FSM, and an action-reviser, is proposed. In the framework, the RL controller learns a Q-function (and control-policy) given a reward function and returns the optimal actions in various situations while the e-FSM develops a Markov driving model. The action-reviser checks the validity of the RL controller's actions using the recently evolved Markov driving model. If the chosen action is invalid, the action-reviser explores and applies an alternative action instead of the RL controller's cho (open full item for complete abstract)

Committee: Umit Ozguner (Advisor); Keith Redmill (Committee Member); Yingbin Liang (Committee Member); Dimitar Filev (Committee Member) Subjects: Artificial Intelligence; Computer Engineering; Electrical Engineering; Robotics

PhD, University of Cincinnati, 2015, Engineering and Applied Science: Aerospace Engineering

Fuzzy Logic Control is a powerful tool that has found great success in a variety of applications. This technique relies less on complex mathematics and more on "expert knowledge" of a system to bring about high-performance, resilient, and efficient control through linguistic classification of inputs and outputs and if-then rules. Genetic Fuzzy Systems (GFSs) remove the need of this expert knowledge and instead rely on a Genetic Algorithm (GA) and have similarly found great success. However, the combination of these methods suffer severely from scalability; the number of rules required to control the system increases exponentially with the number of states the inputs and outputs can take. Therefor GFSs have thus far not been applicable to complex, artificial intelligence type problems. The novel Genetic Fuzzy Tree (GFT) method breaks down complex problems hierarchically, makes sub-decisions when possible, and thus greatly reduces the burden on the GA. This development significantly changes the field of possible applications for GFSs. Within this study, this is demonstrated through applying this technique to a difficult air combat problem. Looking forward to an autonomous Unmanned Combat Aerial Vehicle (UCAV) in the 2030 time-frame, it becomes apparent that the mission, flight, and ground controls will utilize the emerging paradigm of Intelligent Systems (IS); namely, the ability to learn, adapt, exhibit robustness in uncertain situations, “make sense” of the data collected in real-time and extrapolate when faced with scenarios significantly different from those used in training. LETHA (Learning Enhanced Tactical Handling Algorithm) was created to develop intelligent controllers for these advanced unmanned craft as the first GFT. A simulation space referred to as HADES (Hoplological Autonomous Defend and Engage Simulation) was created in which LETHA can train the UCAVs. Equipped with advanced sensors, a limited supply of Self-Defense Missiles (SDMs), (open full item for complete abstract)

Committee: Kelly Cohen Ph.D. (Committee Chair); Corey Schumacher Ph.D. (Committee Member); Elad Kivelevitch Ph.D. (Committee Member); Ali Minai Ph.D. (Committee Member); Grant Schaffner Ph.D. (Committee Member) Subjects: Aerospace Materials

Doctor of Philosophy, The Ohio State University, 2015, Civil Engineering

International construction projects carried out by foreign construction organizations may be subject to various risks due to the nature of the construction industry and business environment of the host country. The regulations of Kuwait oblige foreign organizations to create joint venture relations with qualified local partners to conduct business in Kuwait. A joint venture is a business strategy used to reduce expected risks when expanding into an international market; however, new sources of risk are created via this strategy. This research produced a risk management model for international construction in Kuwait (RIMMICK) designed to facilitate the success of international public construction projects in Kuwait. The aim of this research was to support various governmental agencies such that their projects can be successfully completed. The goal of this work was to reveal and assess the most critical risk factors that can influence the success of international construction projects in Kuwait. RIMMICK is a risk management model designed to identify the risk associated with international construction joint venture (ICJV) projects, assess the identified risk factors and provide project-specific risk ratings, reveal the consequences of the assessed risks, and suggest adequate risk mitigation and response strategies. This model focuses on the most critical phases of a construction project's lifecycle: the bidding and construction phases. RIMMICK can help sustain a risk-controlled environment by continuously monitoring a project's progress and providing reliable risk assessment. The risk intelligent system for construction in Kuwait (RISCK) is a tool designed to facilitate risk evaluation and control processes in RIMMICK. This knowledge-based system was tailored to ICJV projects in Kuwait that are owned by Kuwaiti governmental agencies. The RISCK tool can be utilized during two critical phases of a construction project: bidding and construction. During the bidding ph (open full item for complete abstract)

Committee: Fabian Hadipriono Tan (Advisor); Frank Croft (Committee Member); Rachel Kajfez (Committee Member) Subjects: Civil Engineering

MS, University of Cincinnati, 2014, Engineering and Applied Science: Mechanical Engineering

The objective of this thesis is to develop an application in Enterprise PDM for semi-intelligent part numbering schema using Group Technology, for customers of Dassault Systemes Solidworks Corp. Most of the PDM tools in the market do have a part numbering module, but EPDM does not have one. It is easier to have a sequential part numbering schema, but complexities creep into the system as soon as we decide to go for a significant or intelligent part numbering schema. Also, the huge amount of resources involved in designing an intelligent part numbering system is the major reason why most PDM tools don't prefer to have a part numbering module in them. There is ample research on how sequential part numbering is the perfect approach for companies, but considering the fact that intelligence in part numbers is here to stay for a long time, it is imperative to think of an application that would cater to the needs of the companies, who would like to work with intelligent part numbers. Sequential and intelligent part numbers have their own disadvantages, but working on semi-intelligent part numbers, which give us the best of both worlds, would be the approach that many would prefer. Group Technology deals with item classification and coding techniques to streamline manufacturing processes. These classification and coding techniques in conjunction with semi-intelligent part numbers can turn out to be a novel solution to the problem of unique item identification. This approach should be acceptable to supporters of both sequential and intelligent part numbering. Use of Group Technology adds a different dimension to item identification and in the course of the study proves that this topic needs more research and development to be able to achieve maximum design and manufacturing efficiency in a company. Hopefully, the proposed semi-intelligent part numbering schema fulfills the need of the Dassault Systemes customers interested in this thesis. That would be the scale to mea (open full item for complete abstract)

Committee: Anil Mital Ph.D. P.E. (Committee Chair); Raj Manglik Ph.D. (Committee Member); David Thompson Ph.D. (Committee Member) Subjects: Mechanics

MS, University of Cincinnati, 2022, Engineering and Applied Science: Electrical Engineering

Applied Machine Learning on real-time Cyber Physical Systems (CPS) brings several new challenges to Machine Learning (ML) based control. CPS are subjected to environmental changes, noise, hardware limitations and tightly coupled time constraints, which make real-time control a non-trivial task. This thesis work focus on studying applicability of ML based control in real-time CPS using physical pinball machines as sandboxes. A simulator framework to evaluate ML algorithms in a virtual setting and a real-world framework to evaluate ML algorithms in physical pinball machines are developed. Both frameworks provide visual information and extracted features for the ML agent, and actuates the system according to ML agent control signals. The real-world framework utilizes a real-time state tracker, hardware based synchronizer, and a non-invasive system actuation method to realize the abstracted framework. We discuss the development of the simulation framework and the real-world framework. Subsequently, we move into the application of model-free ML, where we experiment with reinforcement learning under different perception models and modular learning. Finally, we discuss the application of model-based ML where we experiment with Model Predictive Control (MPC) with Deep Neural Networks (DNN) and Support Vector Regression (SVR), on selected primitive goals in the system. Each technique is statistically evaluated and results are presented. The evaluation results showed that ML based MPC was able to reach up to 96% accuracy in the selected shot aiming scenario.

Committee: Zachariah Fuchs Ph.D. (Committee Member); Ali Minai Ph.D. (Committee Member); John Gallagher Ph.D. (Committee Member) Subjects: Electrical Engineering

Doctor of Philosophy, The Ohio State University, 2018, Civil Engineering

Dougong, one of the unique features of ancient Chinese architecture, are located at the intersections of pillars under the roofs in the support systems of ancient Chinese buildings. Virtual reality (VR) is a way to recreate or simulate an environment in which users can interact with objects with high realism and have an immersive experience exploring this virtual world. So far, 38 types of dougong have been modeled in 3-D graphics and imported into the virtual reality environment to establish a complete dougong 3-D graphical library, in which the user can browse and review dougong knowledge with multiple presentation methods in an immersive and interactive experience. Furthermore, a knowledge-based system with the inference engine as a decision tree, the Intelligent Dougong System in Virtual Reality (IDSVR), has been developed as a learning platform to introduce and comprehensively simulate dougong structure and construction. To assess the performance of the application of virtual reality adopted in engineering education, a set of surveys was conducted among the users to collect their feedback on IDSVR. The results obtained from this project prove that the technique of virtual reality is a promising approach to reconstructing ancient buildings and structures such as the Chinese dougong that was modeled, presented, and simulated in this project.

Committee: Fabian Tan (Advisor); Lisa Burris (Committee Member); Abdollah Shafieezadeh (Committee Member); Michael Parke (Committee Member) Subjects: Archaeology; Architecture; Civil Engineering; Computer Engineering; Education; Engineering; History

Master of Science, The Ohio State University, 2014, Civil Engineering

Dougong is a trait of Chinese architecture and specifically the construction engineering of its roofing system, which can be translated into `cap and block' bracket system. In this research, the Song-style dougong recorded in Yingzao Fashi, the oldest currently preserved construction specifications, are systematically reconstructed using 3-D digital graphical techniques. During the modeling process, step-by-step construction operations of more than 20 different kinds of dougong and the delicate workmanship of the components are presented in detail. Then, the categories of Song-style dougong are compiled in a knowledge-based intelligent dougong system (IDS). The IDS presents graphical simulations of the construction process of the Song-style dougong, based on the queries prompted to the users. It allows users to select variables such as dimensions, the number of levels, and the type of components, among others to find the type of dougong expected. Once selection processes are completed, IDS furnishes users with a conclusion concerning the construction process, both textual and graphical, the step-by-step erection of dougong in both isometric view and exploded isometric views from various angles. Through the analyses and observations of Song-style dougong in 3-D graphical environment, the beauty and wisdom of ancient Chinese architecture can be viewed and recognized. In conclusion, the study introduces an important element in ancient Chinese architecture and a new attempt to reconstruct it in digital graphics

Committee: Fabian Tan (Advisor); Frank Croft (Committee Member); Rachel Kajfez (Committee Member) Subjects: Archaeology; Architecture; Art Education; Art History; Artificial Intelligence; Asian Studies; Civil Engineering; Engineering; Wood

PhD, University of Cincinnati, 2000, Engineering : Computer Science and Engineering

A novel Web-based intelligent tutoring system, called Arthur, is developed in this work. This system provides adaptive instruction between various instruction methods that are created by expert tutors. Intelligent tutoring systems are forms of expert systems, where each tutor is an expert in the field and has a different instruction style. Arthur makes use of learning styles theory (Dunn 1978) and mastery learning (Bloom 1976), from education, and case-based reasoning (Kolodner 1993), from artificial intelligence, to bring this new style of asynchronous instruction to the World Wide Web. Case-based reasoning is used to adaptively change instruction methods when corrective instruction is necessary. Unlike the traditional tutoring environment or classroom environment where there is a one-to-one relationship or one-to-many relationshp between the tutor and student or students, Arthur provides a many-to-one relationship between the tutors and student. Imagine taking a course where the student has an unlimited number of tutors available. The purpose of this system is to provide effective instruction via the Web in search of "A Significant Difference" (Russell 1999) in learner outcomes. Chapter 1 gives an introduction to this research. Chapter 2 reviews previous contributions through other Web-based systems that consider learning style as part of their design. Chapter 3 will focus on Arthur's design and explain "What is Arthur?" Chapter 4 discusses the experiment and the experiment results. Finally, Chapter 5 will summarize Arthur and its contributions.

Committee: Chia Han (Advisor) Subjects:

MS, University of Cincinnati, 2010, Engineering and Applied Science: Mechanical Engineering

Robots are getting more and more involved in every facet of life. They have replaced humans in many fields with their dependability and adaptability. While they have been around for more than half a century, only recently have they developed the intelligence to sense their surroundings and behave accordingly. This thesis describes an approach to navigating a robot along a roadway-type course bound with white lines and strewn with obstacles. The Bearcat Cub is the University of Cincinnati's mobile robot developed for competing at the Intelligent Ground Vehicle Competition (IGVC). The Autonomous Challenge component of the IGVC requires the robot to navigate through an unknown complex obstacle course. The robot must sense its surroundings, detect its confines and calculate a heading. The Bearcat Cub utilizes vision and laser sensor reading to detect the lanes and obstacles respectively. This thesis describes how the robot processes the camera images, detects the lanes and calculates a heading. It also discusses how obstacles are detected and suitable avoidance methods. Finally, a new approach to combining both systems has been described. This avoids conflict in the results from both systems. Various lighting conditions and obstacle layouts have been tried and tested to ensure the algorithms effectiveness. The vision processing algorithm discussed in this thesis was tested in the 2008 Intelligent Ground Vehicle Competition with acceptable outcomes. The obstacle avoidance methods have been developed later and can be implemented in future competitions.

Committee: Ernest Hall PhD (Committee Chair); Ronald Huston PhD (Committee Member); Manish Kumar PhD (Committee Member) Subjects: Robots

Master of Science, The Ohio State University, 2010, Electrical and Computer Engineering

This thesis presents an implementation of a Virtual Sensor System which extended the testbed capabilities of the Control and Intelligent Transportation Research Lab at The Ohio State University. Through this system physical objects including robots and obstacles were modeled in a simulation environment and oriented based on their state in the physical world through the use of an existing Virtual Positioning System. Virtual sensors were able to be utilized by robots without the added cost of purchasing the sensor and creating an interface between it and the robot controller. Other components like a traffic light were added in order to improve the visualization abilities of the simulation environment. Three sensors were implemented within this system. A laser range finder was created based on a sensor model already included in the underlying simulator. A new model was implemented to provide measurements from a magnetometer. A fictitious lane edge sensor was also designed to show situations where sensor data can be generated independent of constraints in the physical environment. In order to test the usability of the Virtual Sensor System, two test scenarios were devised. Proportional and proportional-integral controllers were designed independently using the lane edge sensor and a magnetometer array to control a robot as it followed along a simple path. The laser range finder was tested in a small test area containing randomly placed obstacles. Using only the information obtained through this sensor, a control algorithm was written so that the robot maneuvered through the environment without coming into contact with the obstacles. At the same time, the robot produced a map representation of the explored portions of the test area.

Committee: Dr. Ümit Ozgüner (Advisor); Dr. Yuan Zheng (Committee Member) Subjects: Electrical Engineering

Master of Science, The Ohio State University, 2003, Electrical Engineering

In the latest years there have been extensive advances in the field of intelligent vehicle systems and virtual environment simulation technology has played an important role in these advances, enabling researchers to design, develop and test new technologies utilizing computer-based test-beds, without the standard problems associated with real world testing. In this thesis, developments on the Ohio State University Virtual Environment System (OSU-VES) will be presented, which is one such system being developed at The Ohio State University since 1996. First, two main components of the system, which are the Virtual Environment Builder RoadEZ, and the Virtual Environment Simulator VESim, will be discussed in detail. We will present the implementation details of the automation of the environment generation process, including the inputting of the environment data; function fitting to road data; detection and handling of intersections and merges; interpolation, population and triangulation of road, terrain and vehicle path data; placement and editing of 3D objects; and data and configuration generation for virtual simulation. Next we will discuss the advances on the virtual environment simulation process, including key concepts such as modular structure, generic modules for decoupling of module algorithms from simulation interface, triggering events and event trigger modules. This will be followed by three applications of OSU-VES to demonstrate the usefulness and power of the system for intelligent vehicle research: Synchronization of virtual environment simulations with actual sensor data, development of an emergency driver assistance system called the Control Authority Transition (CAT) System, and the building of the environments and scenarios for a human factors research on driver distractions.

Committee: Umit Ozguner (Advisor) Subjects:

Doctor of Philosophy, The Ohio State University, 2007, Educational Theory and Practice

With the rise of technology use in college classrooms, many professors are open to structuring their classrooms in innovative ways. The classroom flip (or inverted classroom) is one such innovative classroom structure that moves the lecture outside the classroom via technology and moves homework and practice with concepts inside the classroom via learning activities. This research compares the classroom flip and the traditional lecture/homework structure in two different college level introductory statistics classrooms. In the classroom flip classroom, an intelligent tutoring system (ITS) was used to deliver the lecture content outside the classroom. Students completed active learning projects in the classroom that often required the use of a spreadsheet computer program to help students work with the concepts in the course. In the lecture/homework classroom, students attended lectures on course content that included PowerPoint slides, and then students practiced with the course concepts by completing homework from their books outside of class. The learning environment and the learning activity in both classrooms are investigated in this study with respect to activity theory and learning environments research. Students were given the College and University Classroom Environment Inventory (CUCEI) to measure both their learning environment preferences and their learning environment experiences. In addition, data were collected via field notes, classroom transcripts, student interviews, student focus groups, researcher journal entries, and student reflections. The quantitative data were analyzed using t-tests and MANOVA, and the qualitative data were analyzed using grounded theory methods. The findings of this research show that classroom flip students were less satisfied with how the structure of the classroom oriented them to the learning tasks in the course. The variety of learning activities in the flipped classroom contributed to an unsettledness among students th (open full item for complete abstract)

Committee: Douglas Owens (Advisor) Subjects: