Master of Science (M.S.), University of Dayton, 2024, Electro-Optics

Light Detection And Ranging (LiDAR) technology continues to gain significance across various industries, including autonomous vehicles, surveying, mapping, and defense. The demand for precise 3D spatial data necessitates active sensing methods. Flash imaging and scanning LiDAR are two ways of achieving direct-detection LiDAR. Flash imaging LiDAR captures the entire scene instantaneously by emitting a single pulse and measuring the return. Scanning LiDAR, on the other hand, operates by directing a focused laser pulse in a controlled pattern, typically through mechanically steered mirrors, and measures the reflected signals at each point. Due to the losses in the optical system and light propagation in the atmosphere, the received light is at a much lower intensity than the emitted. This calls for the demand of sensitive detectors that are able to convert the returned light into a measurable electrical signal. Traditionally, low-light sensing has been achieved using linear mode or Geiger mode avalanche photodiodes (APDs). While linear mode APDs offer amplification akin to low-noise amplifiers, their gain values are often limited, and higher gain variants like those in HgCdTe are costly. Geiger mode APDs, despite their increased sensitivity, operate as switches with a notable dead time. In contrast, the discrete amplification photon detector (DAPD) offers a promising alternative by aiming to achieve single-photon detection without the drawbacks associated with APDs. This study gives comparison between flash and scanning LiDAR systems then focuses on the performance of the DAPD, evaluating the viability of the DAPD for LiDAR applications. As detector technology advances, it not only enhances LiDAR system performance but also broadens its applicability across diverse domains. This research contributes to advancing LiDAR technology, unlocking its potential for even broader adoption and innovation.

Committee: Paul McManamon (Advisor); Andrew Sarangan (Committee Member); David Rabb (Committee Member) Subjects: Electrical Engineering; Electromagnetics; Optics; Physics

Master of Science (M.S.), University of Dayton, 2020, Aerospace Engineering

Velocity measurement inside of a wind tunnel is an extremely useful quantitative data for a multitude of reasons. One major reason is that velocity has a mathematical relationship with dynamic pressure which in turn influences all the aerodynamic forces on the test model. Many devices and methods exist for measuring velocity inside wind tunnels. At the same time, Doppler wind lidar (light detection and ranging) has been used for decades to make air speed measurements outdoors at long ranges. Lidar has been proven effective for many applications, and it has the potential to solve many of the problems faced by current velocimetry techniques inside wind tunnels. Despite this, minimal research has been performed with Doppler wind lidars inside wind tunnels. While multiple commercial systems exist for making air speed measurements at longer ranges, there are currently no widely available commercial devices designed to work well inside wind tunnels. In this research, initial work is described for the design and development of a continuous wave (CW), coherent wind lidar system. The system is for use as an alternative non-intrusive velocimetry method inside wind tunnels relying on the Doppler effect. A scaled down wind lidar designed to operate at much shorter ranges than current commercial wind lidars can be simpler, less expensive, and require less power. A first iteration of the design was constructed for proof of concept testing with a small-scale wind tunnel at low speeds (7.5-9 m/s). Testing showed that the lidar system could take one-dimensional speed measurements of seeded flow that closely matched Pitot static tube data. When not adding tracer particles to the flow, the lidar return signal was not strong enough for the photodetector used to measure the beat frequency. This research is focused on the process for designing the Doppler wind lidar system, constructing the experimental setup, and studying methods for data analysis. Results of testing presente (open full item for complete abstract)

Committee: Sidaard Gunasekaran (Advisor); Aaron Altman (Committee Member); Paul McManamon (Committee Member) Subjects: Aerospace Engineering; Atmosphere; Atmospheric Sciences; Engineering; Optics; Technology

Doctor of Philosophy (Ph.D.), University of Dayton, 2022, Electrical and Computer Engineering

Occlusions are one of the primary data challenges when working with lidar. Unfortunately, occlusions are highly dependent on sensor viewpoints, and efforts to mitigate occlusions involve costly data collection strategies like additional overlap or multiple views. This research focuses on reducing occlusions by generating the missing points in a post-processing step. We introduce an entirely new occlusion dataset for aerial lidar called DALES Viewpoints. We also propose two fundamental changes that we can use in conjunction with current point cloud completion networks to provide an appropriate solution for occlusion reduction in aerial lidar. Specifically, we propose a new method of Eigen feature selection for hierarchical downsampling. This method takes into account point features, in addition to spatial location. We also introduce a point correspondence loss that helps build more robust features by ensuring similar network behavior when processing point clouds that depict the same scene with different physical point locations.

Committee: Vijayan Asari (Committee Chair); David Rabb (Committee Member); Keigo Hirakawa (Committee Member); Theus Aspiras (Committee Member) Subjects: Artificial Intelligence; Computer Science; Statistics

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

Odometry in robotics is the concept of estimating the change in position of a robot with respect to some known position or origin. Precise odometry information can serve as a key component in any autonomous navigation, path planning, and map building applications. It is well known that the position estimate from the wheel odometry sensor is highly noisy due to wheel slippage. This has created the need for using different exteroceptive sensors like sonar, LiDAR, camera, etc. for odometry information. 2-D LiDAR is one of the popular sensors used for odometry estimation in the field of mobile robotics. LiDAR sensor measures time taken for reflection of the laser ray from surroundings to scan the environment. Scan matching algorithms can then be used for finding overlapping information between consecutive LiDAR scans for estimating the pose of the robot. This research proposes an algorithm based on this technique for estimating the pose of a mobile robot. Instead of using raw LiDAR scan, feature-based matching is used in which geometric features are extracted from the scan and matched across consecutive scans for estimating the pose of the robot. The proposed algorithm uses sequential RANdom SAmple Consensus (RANSAC) for extracting line features in the environment and a meta-heuristic Genetic Algorithm for scan matching. Extended Kalman Filter (EKF) has also been implemented for fusing the odometry information from the LiDAR sensor with Inertial Measurement Unit (IMU) sensor to further improve the odometry estimate of the robot. The proposed algorithm has been developed for Robot Operating System (ROS) for real-time processing making it convenient to be implemented on an actual hardware system. The algorithm has been tested in a simulation environment with experiments also performed on an actual LiDAR sensor.

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

Doctor of Philosophy (Ph.D.), University of Dayton, 2021, Electro-Optics

This research presents methods and results of characterizing and correcting PIN photodiode 3D flash LiDAR cameras, with the goal of significantly simplifying and improving the calibration system design. 3D flash LiDAR detectors use time to digital conversion (TDC) circuits to estimate the time of flight of a pulse when a detection threshold is met. As the underlying time to digital conversion (TDC) circuits require more space and power, these circuits will cause, in high bus loading events, electronic crosstalk. These events are more likely to occur in situations where many detectors simultaneously trigger, something that can occur when viewing a flat object head-on with uniform illumination, thus limiting these sensors to image a full frame due to this simultaneous ranging crosstalk noise (SRCN). Solutions were devised including using a windowed region of interest to mitigate additional noise by preventing triggering on all of the focal plane array (FPA) except the windowed region, and methods using a checkerboard pattern for imaging the full frame, including using a physical target downrange and a spatial light modulator.

Committee: Paul McManamon (Committee Chair); Edward Watson (Committee Member); Russell Hardie (Committee Member); Andrew Huntington (Committee Member) Subjects: Optics

Doctor of Philosophy (Ph.D.), University of Dayton, 2019, Electro-Optics

This research demonstrates registration algorithms specific to multi-pixel imaging Inverse Synthetic Aperture LiDAR (ISAL) complex data volumes. Two registration approaches are considered, a mutual information registration algorithm (MIRA) and an enhanced, range bin-summed cross-correlation algorithm. For implementing these in the context of an ISAL signal, a theoretical mapping of the reflected target plane field to an aperture plane for multi-pixel detection is done. The theory for implementing both MIRA and cross-correlation enhancements is detailed and applied to a simulated sensitivity analysis that compares algorithm convergence and performance for different SNR, sub-aperture shift distances, and low pixel supports. To the best of the authors' knowledge, this is the first application of 3D complex volume mutual information registration to LiDAR aperture synthesis. The enhanced cross-correlation algorithm showed significant gain in registration operability with respect to SNR and sub-aperture shift, giving new options for potential ISAL system design. An experimental Flash LiDAR system was constructed utilizing a multi-pixel temporal heterodyne detection approach for simultaneous azimuth, elevation, range and phase ISAL imaging of a target and this system was used to benchmark registration sensitivity for real data volumes. This is the first known application of a fast focal plane array for low support flash temporal heterodyne LiDAR for aperture synthesis.

Committee: Matthew Dierking Ph.D. (Advisor); Partha Banerjee Ph.D. (Committee Member); David Rabb Ph.D. (Committee Member); Bryce Schumm Ph.D. (Committee Member); Edward Watson Ph.D. (Committee Member) Subjects: Electrical Engineering; Optics; Physics

Master of Science (M.S.), University of Dayton, 2018, Electro-Optics

In this thesis, Electro-Optic (EO) range signatures are obtained with a Short-Wave Infrared Super-Continuum Laser (SWIR-SCL) source. 3D printed canonical targets of interest are illuminated by the SWIR-SCL pulsed laser. The scattered laser light from the target is directly detected in mono-static and bi-static configurations with a fast, high bandwidth Indium Gallium Arsenide (InGaAs) PIN photodiode. Temporal pulse returns provide target shape, orientation, and surface roughness information. Spatial and temporal analysis of the collected intensity distribution is performed in MATLAB. Macro and micro surface properties are identified from the collected data by correlating pulse amplitude variations with known range scenes. Finally, range resolution improvement is investigated by means of Tomographic Reconstruction using Radon Transforms and by image processing techniques such as Deconvolution.

Committee: Edward Watson Ph.D. (Advisor); Paul McManamon Ph.D. (Committee Member); Joe Haus Ph.D. (Committee Member) Subjects: Computer Engineering; Electrical Engineering; Engineering; Experiments; Optics; Physics; Scientific Imaging

Master of Science in Electrical Engineering (MSEE), Wright State University, 2017, Electrical Engineering

This thesis explores a novel detection architecture for use in a Direct-Detect Flash LIDAR system. The proposed architecture implements detection of the last two surfaces within single pixels of a target scene. The novel, focal plane integrated detector design allows for detection of objects behind sparse and/or partially reflective covering such as forest canopy. The proposed detector would be duplicated and manufactured on-chip behind each avalanche photodiode within a focal plane array. Analog outputs are used to minimize interference from digital components on the analog input signal. The proposed architecture is a low-footprint solution which requires low computational post-processing. Additionally, constant fraction discrimination is used to mitigate range walk. The proposed architecture is designed in 90nm CMOS technology. The footprint is 170.1 µm² with the largest transistor dimension being 22 µm. The design is easily expandable in hardware to allow additional surfaces to be detected.

Committee: Saiyu Ren Ph.D. (Advisor); Arnab Shaw Ph.D. (Advisor); Ray Siferd Ph.D. (Committee Member); Robert Muse (Other) Subjects: Electrical Engineering

Master of Science (MS), Wright State University, 2016, Earth and Environmental Sciences

The purposes of the following studies were to investigate natural and human influences on several spatial and temporal aspects of a local and regional environment. The decreasing discharge rate of the ground water supplied Yellow Spring may be caused by the increase in population of the nearby Village of Yellow Springs, Ohio. Periodic measurements of Yellow Spring's discharge rate compared to changes in the town's population showed an inverse relationship, where spring discharge declined as population grew. A sharp decrease in discharge occurred during a period when the spring's facade was modified and an airport was built partially overlying the spring's recharge area. These events are believed to have had a greater impact on spring discharge rate than changing population because discharge rate remained relatively constant after its sharp decline, while population began to decline. Aquifer volume change was determined by calculating the volume difference between decadal average water tables that were modeled with ArcMap from water well depth to water measurements and LiDAR elevation data. Counterintuitively, aquifer volume generally increased with population then fell sharply as the population gradually decreased. A slight increase in aquifer volume after withdrawal wells were installed suggests that human consumption had little impact on aquifer volume. When compared to the average Palmer Hydrological Drought Index, aquifer volume generally lowered during dry periods, and rose during wet periods. Minor variations in climate can greatly impact aquifer volume because precipitation only needed to have decreased by 0.26 percent over a 40 year period to account for the lowest calculated aquifer volume. Determining the composition and spatial extent of land uses through land use classification increases our understanding of processes that are harmful to the environment. Because of LiDAR's high spatial resolution, the ability to classify marginally rural land uses (open full item for complete abstract)

Committee: Doyle Watts Ph.D. (Advisor); Songlin Cheng Ph.D. (Committee Member); Abinash Agrawal Ph.D. (Committee Member) Subjects: Environmental Science; Geography; Hydrology; Information Systems; Physical Geography; Remote Sensing

Doctor of Philosophy, The Ohio State University, 2005, Geodetic Science and Surveying

The objective of this research is to reconstruct 3D building models from imagery and LIDAR data. The images used are stereo aerial photographs with known imaging orientation parameters so that 3D ground coordinates can be calculated from conjugate points; and 3D ground objects can be projected to image spaces. To achieve this objective, a method of synthesizing both imagery data and LIDAR data is explored; thus, the advantages of both data sets are utilized to derive 3D building models with a high accuracy. In order to reconstruct complex building models, the polyhedral building model is employed in this research. Correspondingly, the reconstruction method is a data-driven oriented. The general research procedure can be summarized as: a) building detection from LIDAR data; b) 3D building model reconstruction; c) LIDAR data and imagery data co-registration; and d) building model refinement. The main role of aerial image data in this research is to improve the geometric accuracy of a building model. The major contributions of this research lie in four aspects: 1) Two algorithms are developed to perform LIDAR segmentation. Compared with the algorithms proposed by other researchers, these two algorithms work well in urban and suburban areas. In addition, they can keep fine features on the ground; 2) An algorithm of building boundary regularization is proposed in this study. Compared with the commonly used MDL algorithm, it is simple to implement and fast in computation. Longer line segments have larger weights in its adjustment process. This agrees with the fact that longer line segments have more accurate azimuths provided that the accuracy of ending points are the same for all segments; 3) A new method of 3D building model reconstruction from LIDAR data is developed. It is comprised of constructing surface topology, calculating corners from surface intersection, and ordering points of a roof surface in their correct sequence; and 4) A new framework of building model r (open full item for complete abstract)

Committee: Rongxing Li (Advisor) Subjects: Engineering, General

Doctor of Philosophy, The Ohio State University, 2003, Geodetic Science and Surveying

The automatic recognition and reconstruction of buildings from sensory input data is an important research topic with widespread applications in city modeling, urban planning, environmental studies, and elecommunication. This study presents integration methods to increase the level of automation in building recognition and reconstruction. Aerial imagery has been used as a major source in mapping fields and, in recent years, LIDAR data became popular as another type of mapping resource. Regarding their performances, aerial imagery has the ability to delineate object boundaries but omits much of these boundaries during feature extraction. LIDAR data provide direct information about heights of object surfaces but have limitations with respect to boundary localization. Efficient methods to generate building boundary hypotheses and localize object features are described. Such methods use complementary characteristics of two sensors. Graph data structures are used for interpreting surface discontinuities. Buildings are recognized by analyzing contour graphs and modeled with surface patches from LIDAR data. Building model hypotheses are generated as a combination of wing models and are verified by assessing the consistency between corresponding data sets. Experiments using aerial imagery and LIDAR data are presented. Three findings are noted: First, building boundaries are successfully recognized using the proposed contour analysis method. Second, the wing model and hypothesized contours increase the level of automation in building hypothesis generation/verification. Third, the integration of aerial images and LIDAR data enhances the accuracy of reconstructed buildings in the horizontal and vertical directions.

Committee: Anton Schenk (Advisor) Subjects: Geodesy

Master of Science, The Ohio State University, 2024, Civil Engineering

This thesis explores vehicle classification using returns from LiDAR sensors mounted on a moving instrumented probe vehicle (IPV). The first methodology is a pre-existing scheme that classifies vehicles based on height and length, referred to as the height and length method (HLM). The second is a novel scheme that is developed in this thesis that uses the upper envelope of the vehicle's side view silhouette to classify vehicles, referred to as the shape based classification method (SBCM). The pre-existing HLM was developed using stationary sensors and has already been shown to be robust to data imperfections. The present work demonstrates that HLM also works well when the LiDAR sensors are moving. The main limitation of HLM is that it can only achieve a coarse classification: passenger vehicle, single unit truck, multi-unit truck, or vehicle pulling trailer. The novel SBCM is intended to provide finer gradations among classes. The SBCM collects profiles of the vehicle height from LiDAR returns and normalizes the profiles to 100 points equally distanced along the vehicle (i.e., relative percentage distance from the front to the back of the vehicle). A training set of the empirical LiDAR data collected was used to develop prototype vehicle profiles representative of seven classes: passenger car, SUV, mini-van, pickup truck, van, single unit truck and multi-unit truck. Generally, several sub-class height profiles were developed to capture various vehicle shapes within a class. To develop the profiles, vehicles in the sub-class were chosen from concurrent video imagery as being representative of the sub-class. Then the prototype vehicle height profile for the sub-class was determined by taking the average height across all the representative vehicles at each of the 100 points in the normalized profiles. An eighth classification of vehicles pulling trailers is also employed but does not rely on a prototype height profile. Two variants of the 8-class scheme SBCM were a (open full item for complete abstract)

Committee: Benjamin Coifman (Advisor); Rabi Mishalani (Committee Member); Mark McCord (Committee Member) Subjects: Civil Engineering

Master of Science, The Ohio State University, 2024, Food, Agricultural and Biological Engineering

Waste coal in abandoned mine lands poses significant environmental challenges, affecting nearby communities, rivers, and streams. Effective management of these piles is essential due to concerns such as acid mine drainage, soil and water contamination, coal fires, and methane emissions. Various strategies have been proposed for managing waste coal, including potential utilization for rare earth element recovery, soil amendment, construction aggregates, and energy generation. However, the implementation of these strategies remains uncertain due to the lack of precise location and volume data on waste coal piles. Traditional methods for gathering these data rely on field visits and Global Navigation Satellite System surveying, which are costly and labor-intensive. Advances in satellite technologies and the availability of digital elevation models (DEMs) offer an opportunity to estimate waste coal volume on a regional scale in a timely and cost-effective manner. Thus, the objective of this thesis was to develop a robust data analytical framework to locate and estimate the volume of waste coal piles on a regional scale, using the Muskingum River Basin (MRB) in Ohio as the study area. Initially, a prototype was developed to determine the most effective machine learning (ML) model to map waste coal piles in a historical coal mine site within the MRB. While all four ML models effectively identified dominant classes such as Grassland and Forest, the Random Forest (RF) model demonstrated superior performance in classifying the more complex waste coal class, with a precision of 86.15% and recall of 76.71%. Subsequently, the greatest disturbance and reclamation mapping of these waste coal piles were conducted using the LandTrendr algorithm to distinguish waste coal piles in abandoned mine lands from those in active mining areas. Moreover, this study utilized publicly available elevation models to estimate waste coal volume in the MRB. However, since historical terrain mo (open full item for complete abstract)

Committee: Ajay Shah (Advisor); Sami Khanal (Advisor); Tarunjit Singh Butalia (Committee Member) Subjects: Artificial Intelligence; Engineering; Geographic Information Science; Remote Sensing; Sustainability

Master of Science, The Ohio State University, 2024, Environment and Natural Resources

The green salamander (Aneides aeneus) is a unique species of amphibian that specializes in inhabiting the crevices of rock outcrops. These salamanders reach the edge of their range in Ohio, where they are found in two isolated and highly localized populations along the Ohio River. Green salamanders are currently listed as state Endangered; however, an organized status assessment has not been completed in over 15 years. In recent years, the species has experienced notable regional declines across their broader distribution in the Southern Appalachian Mountains. Therefore, it is important to better understand how the salamander's ecology may differ within the fringe populations in Ohio compared to core populations. This project's primary objectives are (1) to update the status and distribution of the green salamander in Ohio, (2) implement light detection and ranging data to investigate and predict which environmental variables influence the species' distribution in the state, and (3) quantify the microhabitat variables affecting the salamander's selection of rock crevices. We hope that such efforts will not only contribute to a better understanding of the ecology and conservation priorities of green salamanders in Ohio but also guide long-term monitoring efforts beyond this study to ensure the future viability of this at-risk species.

Committee: William Peterman (Advisor); Kaiguang Zhao (Committee Member); Stephen Matthews (Committee Member) Subjects: Natural Resource Management

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

Unmanned Aerial Vehicles (UAVs) have seen a rise in applications to various fields. With plenty of algorithms to support automation in UAV flights, Global Positioning System (GPS) is still the major source of position estimation. This has limited the application of UAVs to areas where GPS signal is available and strong. Thus, some other method of position estimation for the UAV is required to expand the UAV application to GPS-denied areas. Moreover, when an operator is piloting a UAV from a remote location, the operator is solely relying on the camera feed coming from the UAV to move the UAV. This camera feed gives a limited field of view of the environment, and the human operator may accidentally run the UAV into an obstacle. In this research, a method of using Hector SLAM for performing position estimation of the UAV in a GPS-denied indoor environment is presented. The Hector SLAM uses a 2D LiDAR mounted on top of the quadcopter to scan the unknown environment. Furthermore, to empower the UAV to autonomously avoid obstacles, an algorithm using Artificial Potential Field method is developed in this thesis which maneuvers the UAV away from obstacles while being piloted by a human operator. The system is developed using Robot Operating System (ROS) and PX4 autopilot. Two different ways, setpoints and attitude commands, of operating the UAV using a keyboard are implemented and compared. The algorithm has been tested in Gazebo Classic simulator and its performance is evaluated.

Committee: Manish Kumar Ph.D. (Committee Chair); David Thompson Ph.D. (Committee Member); Janet Jiaxiang Dong Ph.D. (Committee Member) Subjects: Mechanical Engineering

PhD, University of Cincinnati, 2024, Engineering and Applied Science: Civil Engineering

This research aims at optimizing the winter pavement management practices by (1) developing an efficient yet simple pothole surveillance system utilizing state-of-the-art connected and autonomous vehicle (CAV) sensors, including LiDAR and camera and (2) developing prediction models with machine learning algorithms and applying survival analysis to estimate the service life of commonly used pothole patching methods/materials combinations and ultimately calculating the life cycle cost associated with it. For pothole detection, automobile industry-grade mechanical LiDAR and a camera were used. Two pothole detection approaches were proposed. In the first approach, data collected from the camera and LiDAR were fused and consequently applied deep learning-based object detection algorithm to locate potholes and extract point cloud data corresponding to the potholes using the coordinates of the detected bounding box. The second approach detected potholes directly from the cross-sectional LiDAR point cloud data using a deep learning-based object detection technique followed by utilizing the point clouds' spatial information to estimate the potholes' dimensions. In both approaches, the timestamps of the GPS and LiDAR were synchronized to locate the pothole coordinate. While the first approach showed great promise in estimating the pothole dimensions with unprecedented accuracy, the validation of the second approach demonstrated that the approach can be reliably employed to locate potholes and estimate the dimensions at different highway speeds. Two life cycle cost-based approaches were proposed to optimize pothole patching practice with various combinations of patching materials and methods considering different factors such as traffic volume, size of the potholes, and climatic conditions. The first approach included a non-parametric survival approach to identify the factors affecting the survival life of the pothole patches, consequently using this information to es (open full item for complete abstract)

Committee: Munir Nazzal Ph.D. (Committee Chair); Matthew Steiner Ph.D. (Committee Member); Lei Wang Ph.D. (Committee Member); Nabil Nassif (Committee Member) Subjects: Civil Engineering

Master of Science (MS), Ohio University, 2024, Electrical Engineering (Engineering and Technology)

UAS and UGS, combined with complex sensor suites, are creating opportunities for innovation in mapping and remote sensing. This thesis presents the development and integration of a multispectral suite for the augmentation of airport surface inspections. Multiple host platforms were investigated and a UAS and UGS were selected. A gimbal was used to house all the sensors and supporting hardware. The sensor suite has an integrated GNSS/INS, RTK enabled, navigation source producing high-accuracy PVTA. Specific focus is on the LIDAR sensor and processing of the generated point cloud data using high accuracy georeferencing techniques. The high-accuracy PVTA data from the GNSS/INS was used to generate a georeferenced point cloud using a custom MATLAB script. Additional post-processing including the concatenation of multiple georeferenced point clouds was also performed.

Committee: Chris Bartone (Advisor); David Ingram (Committee Member); Jay Wilhelm (Committee Member); Jundong Liu (Committee Member) Subjects: Electrical Engineering

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

This dissertation explores the potential of utilizing Autonomous Vehicle (AV) technology in the field of transportation asset assessment, inspection, and evaluation. To harness this potential, the study presents a comprehensive framework for developing real-time, lightweight infrastructure evaluation models. As case studies, the dissertation focuses on addressing two critical road and roadside deficiencies: shoulder drop-off and pavement marking retroreflectivity degradation. For each task, data from several road sections is obtained using an AV development platform and standard measuring equipment such as a surveying-grade laser scanner for the first task and a handheld retroreflectometer for the second. Based on the comprehensive field data collection, the study evaluates multiple AI driven approaches for each task. The research demonstrates three automated algorithms for shoulder drop-off assessment using LiDAR data. A method based on moving window filtering and an LSTM neural network exhibited the highest accuracy and best inference time. Additionally, the correlation between pavement marking reflectivity and LiDAR intensity is investigated. A robust end-to-end AI solution is proposed for automated marking extraction and retroreflectivity prediction. The proposed solution is finally evaluated on its robustness to driving speed, scanning lane and direction, and wet road conditions.

Committee: Munir Nazzal Ph.D. (Committee Chair); Donghoon Kim Ph.D. (Committee Member); Lei Wang Ph.D. (Committee Member); Ali Minai Ph.D. (Committee Member); Manish Kumar Ph.D. (Committee Member) Subjects: Artificial Intelligence

Master of Science, The Ohio State University, 2023, Environment and Natural Resources

Urban forests are important infrastructure in cities that hope to mitigate the worst effects of urbanization and climate change. Trees are shown to remove pollution, reduce surface temperature, intercept stormwater, sequester carbon, and secure other ecosystem services (Berland et al., 2017; Escobedo & Nowak, 2009; D. J. Nowak et al., 2008, 2013; Rahman et al., 2018; Speak et al., 2020; Xiao & McPherson, 2002). These beneficial forest processes can be modeled and quantified using environmental conditions and tree characteristics (Bodnaruk et al., 2017; D. J. Nowak et al., 2008; D. J. Nowak, 2021; Rotzer et al., 2019). Among these characteristics, crown structure and leaf metrics are important factors to be quantified in efforts to estimate ecosystem services provided by urban forests (Rotzer et al., 2021a). Because crown structure and leaf metrics are so impactful for estimating ecosystem services, the accuracy of their assessments is important. However, methods to measure these characteristics exist under different levels of assumption, generalization, and uncertainty. For example, the highly heterogenous, fragmented urban forest is subject to unique anthropogenic conditions that can make cities within the same climate region distinct with regards to crown structure and leaf metrics. Such variation makes application of more generalized allometric equations, which describe how the characteristics of living trees change with size, uncertain (Berland, 2020; McHale et al., 2009a). Climate change will only exacerbate this uncertainty by making older allometric models unreliable (Pretzsch et al., 2017). In addition to the uncertainty in model application, tree crown measurements and estimation techniques often differ in dimensionality. For example, allometric models may only characterize a tree crown across its height and width while photogrammetric reconstruction can be a complex mesh made up of thousands of polygons. This difference in a model's dimensionality h (open full item for complete abstract)

Committee: Steve Lyon (Advisor); Yanlan Liu (Committee Member); Matt Lewis (Committee Member) Subjects: Environmental Science; Environmental Studies; Geographic Information Science

Master of Science (M.S.), University of Dayton, 2023, Electro-Optics

Today, with the development of artificial intelligence and autonomous driving in full swing, lidar is playing a vital role. As an important sensing and detection component, lidar uses 3D point cloud images as a medium to allow artificial intelligence systems to perceive the outside world and perform reasoning work. Therefore, the processing and operation implementation of point cloud is an important part of the information processing of a lidar system, which will determine the accuracy and feasibility of artificial intelligence judgment. In this thesis, an analysis method based on extracting point cloud point separation distance distribution features is used. First, we will introduce how a lidar system works and how a lidar system collects information and generates a 3D point cloud. Afterward, feature analysis of point cloud point separation distribution for dimensionality reduction will be proposed. At the same time, we will use the point separation distribution feature to do object classification, object recognition and segmentation of whether there are vehicles on the road. What's more worth mentioning is that we also provide deep learning results and analysis based on point cloud point separation distribution features. On this basis, we discuss the significance and practicality of this feature analysis.

Committee: Edward Watson (Committee Chair); Partha Banerjee (Committee Member); Miranda van Iersel (Committee Member) Subjects: Electrical Engineering; Engineering; Optics