eScholar

eScholar stores, preserves and disseminates digital copies of the research and scholarly output of eScholar faculty, researchers and students. These can include the following items:

  • Monographs
  • Pre- and post-prints of academic journal articles
  • Theses and dissertations
  • Major projects and papers
  • Reports/working papers and conference proceedings

Materials in eScholar are openly available to the world and discoverable through search engines such as Google Scholar. This high visibility, discoverability, and exposure can lead to increased citation. Contact Library Publishing for more information.

Communities in eScholar

Select a community to browse its collections.

Now showing 1 - 5 of 9

Recent Submissions

  • Thumbnail Image
    Item type: Item , Access status: Open Access ,
    Experimental characterization of the dynamic response of a mock-up CANDU fuel bundle
    (2023-12-01) Saudy, Ali; Mohany, Atef
    This thesis discusses the vibration response of a mock-up CANDU 37M fuel bundle. The experiments are performed using non-invasive methods of vibration measurement by means of high-speed cameras and laser displacement sensors. The contact dynamics between the fuel bundle and the pressure tube are replicated using a high precision custom-made setup. Also, the contact between the fuel bundle and the neighboring fuel bundles are replicated using dummy endplates. The results show that the fundamental frequency of the fuel elements changes depending on the contact conditions with the neighboring fuel bundles. Furthermore, the endplates exhibit a complex vibration response with multiple dominant vibration frequencies that include the fundamental frequency of the fuel rods. The results of this experimental work provide a consistent and accurate characterization of the damping ratios and mode shapes of the fuel rods and the whole bundle including the endplates.
  • Thumbnail Image
    Item type: Item , Access status: Open Access ,
    A comparison of digital human model ergonomics outputs driven by optical and inertial motion capture systems in virtual reality
    (2024-12-01) Dharmaputra, Ryuta; La Delfa, Nicholas
    Digital human models (DHM), or virtual human avatars, can be used within digital virtual manufacturing to conduct proactive ergonomics analyses. Recent technological advancements have allowed for dynamic posturing of the DHM using motion capture technology and virtual reality immersion, providing a potentially powerful approach to evaluate ergonomics outputs including hand location and spine compression. The purpose of this thesis was to compare the hand locations, joint angles and spine compressions produced by a DHM software program when driven by three different types of motion capture systems during a virtual reality task. Findings show that the inertial-driven model consistently underestimated hand locations by 24.3 cm, whereas the Vive trackers showed an overall RMS error of 13.7 cm. This error is further supported by spine compression values, where IMU-driven DHMs underestimated up to 213N compared to the optical system. As a result, the DHM generated using kinematic data from the Vive Trackers system provided the most similar results to the actual location and represent a promising, low cost solution for future ergonomics analyses.
  • Thumbnail Image
    Item type: Item , Access status: Open Access ,
    Design and development of an autonomous mobile robot for parking lot mapping
    (2026-02-01) Selvanathan, Jonathan; Lang, Haoxiang
    Autonomous mobile robots have become a viable solution for advancing parking lot infrastructures to be autonomously patrolled and surveyed remotely. Towards this goal, this thesis presents a monocular vision-based perception system that generates a visually interpretable parking lot map using a constructed 1/3rd scale Security Patrol Robot (SPR). The proposed perception pipeline uses semantic segmentation, inverse perspective mapping, probabilistic line detection, and pose-based line clustering to approximate metric-scaled line marking representations. An incremental mapping algorithm uses robot localization to merge spatially related line detections and create a globally consistent map. Parking spots are detected based on mapped geometry and occupancy is classified using 3D vehicle detection. The end-to-end system is verified using data captured by the SPR in an outdoor parking lot environment. The results show the developed framework is capable of efficient parking lot mapping and spot detection under real-world conditions.
  • Thumbnail Image
    Item type: Item , Access status: Open Access ,
    Multi-server federated learning in vehicular edge computing
    (2026-03-01) Mazloomi, Fateme; Heydari, Shahram S.; El-Khatib, Khalil
    Federated learning (FL) offers a promising paradigm for privacy-preserving model training in connected and autonomous vehicle networks, where vehicles act as clients and roadside units (RSUs) host FL servers at the edge. However, practical deployments face multiple challenges: highly non-IID and noisy data across vehicles, heterogeneous computation and communication resources, private training costs, intermittent connectivity, and the need to coordinate multiple servers. This thesis addresses these challenges through three complementary contributions. First, we propose T-BIDS, a dynamic, budget-aware client selection framework that explicitly accounts for client quality, cost, and availability. T-BIDS uses Thompson sampling to estimate each client’s contribution quality from validation improvements, and in each round selects participants by a quality-to-bid ratio under a per-round budget, with incentive-compatible payment rules. Clients can also augment their local datasets with targeted synthetic samples to rebalance under-represented classes. Across extensive simulations on CNNs on MNIST and GTSRB under shard- and Dirichlet-based non-IID splits with client-level label noise and mobility-induced dropouts, as well as additional experiments with ResNet-18 on GTSRB, T-BIDS converges faster and achieves higher, more stable accuracy and lower loss than baseline selectors. Second, we design an edge-based multi-server FL framework that combines performance-aware aggregation with mobility support. Each server aggregates peer updates using statistical weighting and outlier mitigation, and an application-layer handover protocol preserves model updates when vehicles move between RSU coverage areas. Compared with single-server and edge–cloud baselines, the proposed multi-server methods achieve higher accuracy and improved precision, recall, and F1-score while maintaining low latency and avoiding the additional model-transfer delays of cloud-based aggregation. Third, we propose a client-feedback-driven, quality-based reliability mechanism for model sharing in multi-server FL. In this approach, neighboring servers exchange global models, and client-side evaluations are aggregated to maintain per-peer reliability scores. These scores guide which peer model(s) are selected for reuse or dissemination as the next round initialization, enabling more effective knowledge transfer across servers—particularly in the early rounds—without requiring raw data exchange or explicit disclosure of sensitive client dataset statistics. Overall, the thesis demonstrates that coordinating clients and servers based on realtime quality, cost, and reliability yields robust, scalable, and privacy-preserving FL for vehicular edge computing.
  • Thumbnail Image
    Item type: Item , Access status: Open Access ,
    On development and perception of theomorphic robots: the uncanny valley effect to Artificial Intelligence (AI)
    (2026-01-01) Carrasco Garcia, Carlos Alberto; Hung, Patrick; Vargas Martin, Miguel
    This thesis presents the development and evaluation of a theomorphic robot, the Korean Pensive Buddha Robot, a life-sized, articulated robotic statue modelled after the Korean National Treasure No. 83, the Pensive Bodhisattva. The project integrates three-dimensional (3D) printing, servomotor actuation, and conversational Artificial Intelligence (AI) to create a robot capable of symbolic movement and interaction. While initially conceived as a platform for Human–Robot Interaction (HRI) studies, this research adopts a novel methodological shift: rather than directly testing with human participants, the evaluation uses OpenAI’s Contrastive Language–Image Pre-training (CLIP) model to simulate perception and classification processes. Specifically, the research uses the conceptual structure of the Godspeed Questionnaire, widely used in research on the Uncanny Valley effect, to investigate how an AI system categorizes a robot relative to other entities. The Uncanny Valley, introduced by Masahiro Mori (1970), describes how human affinity toward anthropomorphic robots and avatars increases with human likeness until subtle imperfections trigger discomfort or eeriness, causing a sharp drop in emotional acceptance. This phenomenon produces the characteristic “valley” between realism and affective response. CLIP is tasked with evaluating images of four categories: the Pensive Buddha Robot, its first half-scale prototype, a still statue, and a real human. The model’s outputs are analyzed across Godspeed dimensions of anthropomorphism, animacy, likability, perceived intelligence, and safety. By applying an AI-based lens to questions traditionally addressed through human evaluation, this research also investigates whether machine perception aligns with the Uncanny Valley effect. The results provide a new perspective on how AI vision systems interpret robotic embodiment and symbolic form, while also testing the applicability of social robotics assessment tools in computational contexts. Ultimately, this research contributes to both cultural robotics and methodological innovation in evaluating robot perception, without relying exclusively on human subjects.