Browsing by Author "Hu, Yaoping"
Now showing 1 - 20 of 30
Results Per Page
Sort Options
- ItemOpen AccessA study on effect of a haptic device on performing a robot-assisted surgical task(2011) Samad, Manar Din; Hu, Yaoping; Sutherland, Garnette R.
- ItemOpen AccessAlgorithms for segmentation of lung lobes in volumetric ct images for surgical planning of treating lung cancer(2007) Wei, Qiao; Hu, Yaoping
- ItemOpen AccessApproach to ease design criteria of a real-time model used in a VR training system considering constraints of human perception(2012-07-23) Widmer, Antoine; Hu, YaopingVirtual reality (VR) systems have potential of contributing to the training of medical students in a variety of procedures. This thesis focuses on a design issue related to developing VR training systems for soft tissue (e.g., breast phantom) palpation. In such a VR system, it is paramount to provide a real-time model that simulates physical behavior of an actual breast phantom. However, it is difficult to design such a real-time model with high accuracy due to time and physical constraints. To mitigate this difficulty, I consider constraints of human perception which is insensitive to small discrepancies of objects during real-time interaction. Such consideration could aid to relax design criteria of the real-time model by achieving its accuracy at a certain degree while keeping human perception of object softness unchanged. Therefore, I take a two-step approach to determine visual and haptic (pertinent to force feedback) discrepancies tolerable for this human perception. In the first step, an evaluation method is developed to compute discrepancies of the real-time model for visual displacement and force feedback, compared to its finite element method counterpart featuring physical parameters of a breast phantom. The computation uses statistical analyses which like human perception are insensitive to small discrepancies of datasets. In the second step, two studies are performed to examine the constraints of human perception. The first study reexamined raw data from my MSc work to understand the effect of three popular alignments between a visual display and a haptic device on the human perception of object softness. This study serves to select an alignment producing the least perceptual illusion and physical workload for palpation. Using the evaluation method and the selected alignment, the second study investigates the effect of different discrepancies on the human perception of object softness. It is observed that this perception is insensitive to small discrepancies up to a threshold of 11.0% and 6.3% for visual displacement and force feedback, respectively. This indicates that a real-time model yielding discrepancies of visual displacement and force feedback below their respective thresholds could be sufficient for simulating a soft tissue (such as a breast phantom) during palpation.
- ItemOpen AccessApproaches to Reduce Clutter and Enhance Robustness of Vortex Extraction in Flow Visualization(2017) Padmesh, Kavya; Hu, Yaoping; Smith, Mike; Murari, KartikeyaOver the past few decades, extraction and visualization of flow features like vortices has gained tremendous importance and is employed in numerous applications. Several vortex detectors are available in literature that can identify vortices in most empirical and computational datasets. However, despite these efforts, uncertainties in empirical measurements often results in undesired vectors that cause clutter in visualization. Clutter would obscure vortex features and make it hard to understand complex flow behavior. Additionally, floating-point errors in vortex detector computations lead to false positives in vortex extraction. This thesis aims to solve aforementioned problems by implementing - a pre-processing technique to filter undesired vectors from empirical data and a threshold estimation technique to reduce the effect of floating-point errors in vortex extraction. Proposed methodologies are tested on several flow datasets of various sizes and turbulence intensities. Results indicate enhanced visualization by reducing clutter; also, they confirm improved robustness in vortex extraction.
- ItemOpen AccessCerebellar Model Articulation Controllers for Bilateral Teleoperation with Elastic-Joint Manipulators and Haptic Feedback(2016) L'Orsa, Rachael; Macnab, Chris; Goldsmith, Peter; Hu, Yaoping; Nowicki, EdTelesurgical systems that incorporate haptic feedback at the master with a light, compliant slave may provide a superior surgical experience for both patient and health care team. This thesis presents a first investigation of the use of Cerebellar Model Articulation Controllers in just such an application, where the adaptive neural network is used to overcome unknown, nonlinear system dynamics, interaction with unstructured environments, and non-passive operator behaviour in real-time. The controller's output is shown to be Ultimately Uniformly Bounded via the Lyapunov approach, and its stability is verified experimentally. It performs as well as a PID controller during direct real-time bilateral teleoperation and is more robust to rate changes during hard contacts. An auxiliary error that provides torque tracking during interaction and velocity tracking during motion through free space without the need for switching or gain scheduling is evaluated, and key considerations for the implementation of custom controllers with Quanser's 2 DOF Serial Flexible-Joint Robot are discussed.
- ItemOpen AccessCoherent Dynamics and Energetics in Thin Flat Plate Wakes(2021-08) Agrey, Kaden; Martinuzzi, Robert; Hu, Yaoping; Johansen, Craig; Wan, Richard; Pieper, JeffThe dynamics and energetics of two mean two-dimensional wakes behind a flow-normal thin flat plate, induced by inclusion and exclusion of end plates, are studied. Both wakes are characterized by quasi-periodic vortex shedding but differ in mean topology and typical wake characteristics, such as mean base pressure and recirculation length. Energetically optimal proper orthogonal decomposition modes are used to approximate the coherent motion and thus triply decompose the velocity field into a mean, coherent, and residual field. This is utilized to obtain a dynamic characterization of the wakes and to study the large scale coherent structures and their energetic exchanges with other scales of motion. From this, a slow-varying base flow and lateral shear layer flapping are shown to influence the shedding dynamics differently in each wake. These differences in the mean field and coherent dynamics are related to wake turbulence levels and vortex deformation, and thus are at the beginning of the energy cascade and the energy transfer process related to the wake turbulence levels.
- ItemOpen AccessComputational Modeling of Electrical Cell-to-cell Interactions in Cardiac Tissue: Applications to Model Parameter Selection and Pacemaker Function(2017) Kaur, Jaspreet; Vigmond, Edward; Nygren, Anders; Di Martino, Elena; Hu, Yaoping; Murari, Kartikeya; Clancy, ColleenCell-to-cell interactions are important in determining the electrophysiological behavior of cardiac tissue. In this research, computer modeling is used to investigate the importance of these interactions in two different contexts: 1) how to adjust parameters in single cell models to accurately reproduce tissue behavior, and 2) determining requirements for successful conduction at the interface between different tissue types, specifically from the sinoatrial node (SAN) to the atrium. Membrane resistance (Rm), the inverse of the slope of the current-voltage (I/V) relationship for a cardiac myocyte, is an important determinant of electrical cell-to-cell interactions. Experimentally, Rm can be measured by applying a small current and measuring the resulting change in membrane voltage. To investigate the importance of Rm, a multi-objective genetic algorithm approach was developed for enhancing the fitting of action potentials (APs) in single cell models. Rm was fit at several points during the AP along with AP morphology. The results demonstrate that including Rm as a fitting criterion yields improved convergence, reduced variability in parameter estimates, and improved robustness, while specifically improving the ability of the model to reproduce tissue behavior. Bioengineered pacemakers are cellular constructs intended to replace the SAN pacemaker function. The interface between the SAN and atrium appears to have features designed to facilitate conduction. Depending on the species, these features involve gradual transitions (gradients) in ion channel densities and coupling conductance, or insulating boundaries with conduction at discrete exit points only. We used simulations to determine the importance of each of these features, with the aim to provide guidance to future development of bioengineered pacemakers. We found that gradients in ionic conductance (specifically ICaL) are required in rabbit SAN. There is narrow range of coupling for which the SAN is able to propagate towards atrium without coupling gradients. In canine SAN, these gradients support conduction. However, gradients are not required, provided conduction from SAN to atrium is restricted to discrete exit points. This suggests two possible strategies for successful conduction at the interface between a bioengineered pacemaker and the atrium: 1) engineer the construct to have appropriate ionic current and intercellular coupling gradients, or 2) functionally insulate a homogeneous construct from the atrium with conduction only at discrete points.
- ItemOpen AccessComputer-assisted Screening of Motion Artefact for Quality Control in Large-scale MR Imaging Trials(2017) Adair, David; Frayne, Richard; Gobbi, David; Hu, Yaoping; Krishnamurthy, DiwakarAs the scale of medical imaging trials increases, manual quality control of the enormous volume of imaging data becomes intractable and costly. Machine learning may provide solutions to reduce the challenge of these large trials through the development of computer-assisted screening tools. The objective of this dissertation was to evaluate the suitability of machine learning for solving scalability problems of manual quality control by training an automated classifier to detect simulated motion artefact on otherwise high-quality magnetic resonance images of healthy human brain. The classifier achieved high accuracy (98.5%) without any performance optimization, and, incidentally, discovered a screening error within the experiment dataset, further demonstrating the power of machine learning in this domain and encouraging further research towards computer-assisted screening tools.
- ItemOpen AccessEvolution of Large-Scale Structures in the Wake of Sharp-Edge Thin Flat Bodies(2016-01-29) Hemmati, Arman; Wood, David Howe; Smits, Alexander J.; Hu, Yaoping; Martinuzzi, Robert J.; Johansen, Craig T.This thesis describes computational fluid dynamic study of the wake behind thin flat plates at Reynolds numbers large enough for the formation of energetic structures and turbulence. The dynamically rich behavior of unsteady turbulent wake of bluff bodies consists of energetic and large-scale structures generated through flow instabilities, which have an anisotropic and geometry dependent topology. Large eddies are most important in characterizing the wake and provide the largest contribution to kinetic energy. The three-dimensional wake of thin flat plates positioned normal to a uniform flow is evaluated using Direct Numerical Simulations and Large Eddy Simulations. The flow around a 2D plate is examined at Re = 1200 and 2400 to characterize the wake and establish the dynamics of vortex formation and detachment processes. This is extended to the wake of finite aspect ratio (3D) thin flat plates at Re = 1200. The aspect ratios investigated are 3.2, 1.6 and 1.0. Flow topology eduction is carried out by examining the temporal evolution of aerodynamic forces and their phase-angles, as well as velocity and vorticity fields. Large-scale structures are investigated based on their topology, contributions to turbulent kinetic energy, and interaction with the surface pressure. The educed structures in the wake of 2D plates belong to three distinct regimes (H for high-, L for low-, and M for moderate-intensity vortex shedding) determined from periodicity of vortex shedding based on lift and drag fluctuations. The characteristics of previously identified H and L regimes were quantified, while introducing a new regime M. Formation and distortion of spanwise vortex rollers and streamwise vortex ribs coupled with Reynolds stress anisotropy and compression or stretching of the recirculation region characterize main differences among these regimes. The introduction of additional shear layers significantly alters the flow topology and vortex shedding process for 3D plates compared to 2D plates. Vortices are formed on longer edges of the plate, whereas shear layers on the shorter sides are carried away by the induced streamwise flow. This results in a single vortex shedding process. The vortex “peel-off” on shorter edges fixes the vortex detachment at sharp corners of the plate.
- ItemOpen AccessEvolution Surfaces for Spatiotemporal Visualization of Vortex Features(2019-09-10) Ferrari, Simon; Hu, Yaoping; Martinuzzi, Robert JohnTurbulent fluid flow data is often 4-dimensional (4D), spatially and temporally complex, and requires specific techniques for visualization. Common visualization techniques neglect the temporal aspect of the data, limiting the ability to convey feature motion or offering the user a complicated visualization. To remedy this, we present an approach – evolution surfaces – focused on the spatiotemporal rendering of user-selected flow features (i.e., vortices). By abstracting the spatial representation of these features, the approach renders their spatiotemporal behavior with reduced visual complexity. The behavior of vortex features are presented as surfaces, with textures indicating properties of motion and evolution events (e.g., bifurcation and amalgamation) represented by the surface topology. We evaluated the approach on two datasets generated from empirical measurement and computational simulation (Re = 28000 and Re = 1200 respectively). Our approach’s focus on handling evolution events makes it capable of visualizing higher Reynolds number (Re) flows than other surface-based techniques. This approach has been assessed by fluid dynamicists to assert the validity for flow analysis. Evolution surfaces offer a compact visualization of spatiotemporal vortex behaviors, opening potential avenues for exploration and analysis of fluid flows.
- ItemOpen AccessExploration on Just Noticeable Difference of Amplitude for Surface Shape Perception in a VE through Haptic Interactions(2021-11-25) Huang, Jing; Hu, Yaoping; Fear, Elise C.; Krishnamurthy, DiwakarHuman ability to perceive surface shapes plays a crucial role in understanding their surrounding objects and environments. Although the sense of vision has been employed frequently for virtual shape perception, it is desirable to enhance performance of shape perception in the three-dimensional (3D) virtual environments (VEs) by providing haptic feedback. To explore how well the human ability is to discriminate surface shapes under haptic feedback, the experiment presented in this thesis applied a novel force model, which was created by our research lab, to distinguish four kinds of sinusoidal virtual surfaces, which have various amplitudes differing from a flat surface. The main finding is that Just Noticeable Difference (JND) range of amplitude was around 1.19 - 1.95 mm for distinguishing sinusoidal virtual surfaces from flat ones. This finding quantifies the human ability to discriminate virtual surfaces, and may contribute to fundamental knowledge for creating applications with haptic interactions.
- ItemOpen AccessFeasibility of Mapping Brain Activity to the Levels of Task Complexity within Environments of Virtual Reality(2023-09-21) Perez Vite, Yobbahim Javier Israel; Hu, Yaoping; Fear, Elise; Tan, BenjaminMapping brain activity to certain levels of task complexity is essential for creating environments of Virtual Reality (VR), which could adapt to the mental states of human users. To investigate the feasibility of such mapping, the research work of this thesis took an approach of two steps. At first, the levels of task complexity were defined according to the geometric and appearance parameters of objects that the users interacted with for executing a task. By associating the parameters to the execution of the task, this step remedied qualitative descriptions of the levels in current state-of-the-art. Secondly, an empirical study of two experiments was conducted within a VR to collect brain activities (as brainwaves) of human participants (i.e., users) during the execution involving various task complexity. Using a device of encephalography (EEG) to collect the brainwaves, this step assessed several existing features derived from the brainwaves as potential indicators of feasibility. This thesis produced two significant findings: (1) the definition of task complexity is quantitative and could be suitable for describing object-oriented tasks, and (2) specific EEG features – such as engagement ratio – could indicate increased or decreased levels of task complexity. Hence, the work indicates the feasibility of mapping brain activity to the levels of task complexity. Future investigations are needed to refine the definition, and EEG features for optimizing cognitive engagement and performance by modulating the levels of task complexity. The outcomes of the investigations could have implications for training, simulation, and user experience in various VR-based applications.
- ItemOpen AccessFramework of multiuser satisfaction for assessing interaction models within collaborative virtual environments(IEEE Press, 2017) Erfanian, Aida; Hu, Yaoping; Zeng, TaoCollaborative virtual environments (VEs) require interaction models for resolving conflicts and promoting multi-user collaboration. Common models, such as the first-come-first-serve (FCFS) model, which grants interaction opportunities to the most agile user, and the static priority model, which gives interaction opportunities to the user with the highest predefined priority, disregard the importance of perceiving equality in interaction (EII) among all users. One exception is the dynamic priority (DP) model, as proposed in our earlier work, which grants interaction opportunities to a user based on the recency of his/her gained opportunities. To date, few research efforts have investigated the effect of interaction models on multi-user satisfaction. This paper hence presents an assessment of the DP model’s effect on multi-user satisfaction within a collaborative VE. We first verified that the DP model allowed multiple users to perceive EII. We then conducted an experiment to examine the effect of the DP and FCFS models on multi-user satisfaction under a quasi-practical scenario that mimicked a decision-making meeting of experts. The framework of the examination was based on several metrics, which we proposed for the components of the ISO/IEC 25010:2011 standard. This framework resolved issues with existing metrics that measure user satisfaction by analyzing individual experience, thus omitting EII desired by multiple users. The results of the experiment indicated that the DP model fulfilled the metrics of the framework significantly better than the FCFS model. This observation implies a potential application of the DP model in collaborative VEs where multi-user satisfaction is the key to productive collaboration.
- ItemOpen AccessIdentifying the Problems of Software Re-architecting and a Knowledge Representation Framework to Address Them(2018-06-25) Moazzen, Elham; Walker, Robert J.; Denzinger, Jörg; Oehlberg, Lora A.; Anvik, John; Hu, YaopingReal-world software undergoes constant change: to fix bugs; to extend functionality; to interact with the changing “ecosystem” around it; and to make internal improvements. Non-trivial software must possess a software architecture: a division into smaller pieces, how those pieces are meant to interact, and how those pieces are deployed physically. As a software architecture can have a significant impact on important properties of the software, the architecture for a software system may need to change as the system itself undergoes change: this is software re-architecting. Unfortunately, software re-architecting is poorly understood: without understanding what is involved in software re-architecting and what problems people encounter in approaching it, we cannot help solve or avoid those problems. I begin this thesis by conducting a case study on a real world example of a software re-architecting, for which documentation and records of discussions were available, to find basic issues that arose during the process. I also conducted a series of interviews with software engineers centred around those issues to deepen our understanding of the process of software re-architecting and discovered the notion of discrete change steps that must be organized and coordinated. I identify a set of critical challenges that must be addressed in any concrete solution. Software engineers lacked a systematic approach to the communication and record management of change steps, suggesting a set of design guidelines for future collaboration tools tailored for re-architecting. They need collaboration tools that facilitate viewing, recording, and retrieving the change steps, and involving the communications within and between the levels of the development team. I then propose a knowledge representation framework for the change process in asynchronous collaboration. This framework is a first step toward a re-architecting collaboration tool that would help to systematize the change process without disrupting it. I developed a paper prototype of the framework and conducted a user evaluation study to determine if the new approach meets the needs of software engineers working on a software re-architecting. My study suggests that the ii approach supported by the prototype allows software engineers to better present changes to their team relative to traditional mechanisms, thereby enabling them to consider more detail. I illustrate the potential value of the framework as a platform for deeper study and further investment in tools, highlighting promising areas for future research.
- ItemOpen AccessMechanisms of Integrating Vibrotactile and Force Cues for 3D User Interaction within Virtual Environments(2018-08-29) Tarng, Stanley; Hu, Yaoping; Martinuzzi, Robert; Fapojuwo, Abraham O.; Nielsen, JorgenA model for the mechanism of integration for vibrotactile and force cues in the haptic modality is important to facilitate task performance of human users in a three-dimensional (3D) virtual environment (VE). To investigate this mechanism, the research in this thesis used maximum likelihood estimation (MLE) and proposed proportional likelihood estimation (PLE) as the models of the integration. Two significant findings are as follows: (1) I found that based on the task accuracy, MLE was unable to integrate vibrotactile and force cues despite its ability to integrate cues of different modalities in literature; (2) Integration with PLE revealed that the mechanism of integration of vibrotactile and force cues may not be entirely additive as assumed in MLE. This work sheds an insight for proper model of integration between vibrotactile and force cues for interactive tasks in VEs.
- ItemOpen AccessMethodology of Robot-Assisted Tool Manipulation for Virtual Reality Based Dissection(2019-03-29) Trejo Torres, Fernando Javier; Hu, Yaoping; Sesay, Abu B.; Westwick, David T.; Chan, Sonny; Liu, Peter J.Robot-assisted (RA) surgery employs a master-slave system, in which a surgeon's hand manoeuvres the stylus of a hand controller (master) mapped at the operation site to indirectly manipulate a surgical tool attached to the end-effector of a robot (slave). Hence, RA surgery has two drawbacks. Firstly, the transfer of tool-tissue interaction forces to a surgeon is either absent or inaccurate. Secondly, RA surgery incorporates motion coupling (MC) and motion coupling plus orientation match (MC+OM) as indirect modes of tool manipulation, which disregard a pose (position and orientation) match (PM) between the mapped stylus and the tool. This may cause inadvertent tissue trauma during tasks like dissection, which spends ~35.0% of surgery time. Due to the potential of virtual reality (VR) based surgical training, this thesis presents a methodology to address the drawbacks on a VR simulator of soft-tissue dissection. The methodology comprises the formulations and evaluations of an analytic model that estimates dissection forces; and a PM algorithm. The simulator interfaced with the haptic device PHANToM Premium 1.5/6DOF (as a hand controller) to deliver the model forces, and incorporated the kinematics of the device and neuroArm (a neurosurgery robot) for the PM algorithm. The evaluation of the model for estimating dissection forces collected at the tool speeds of 0.10, 1.27, and 2.54 cm/s indicated a force estimation > 80.0%, a computation time < 1.0 ms (the device's update period), and a bandwidth < 30.0 Hz (the device's bandwidth). Moreover, the model lessened cognitive workload for dissections executed at 0.10 cm/s. The evaluation of the PM algorithm revealed a position match < 30.0 µm (the position resolution of the device and neuroArm), an orientation match < 10.0° (to minimize the surgeon's disorientation), and a computation time < 500.0 µs (a half of the device's update period). Additionally, the algorithm became useful to maintain an accurate tool speed and reduce tissue trauma for dissections performed at 0.10 cm/s. The outcomes imply the suitability of the methodology for VR-based RA dissection and their potential to suggest guidelines for VR-based RA dissection training.
- ItemOpen AccessMulti-objective optimization using evolutionary algorithms: Application to the control of flow past a circular cylinder(2018-11-22) Bingham, Conrad Cole; Martinuzzi, Robert John; Morton, Chris R.; Hu, Yaoping; Ziadé, Paul; Westwick, David T.; Epstein, Marcelo D.Modifications to the vortex shedding dynamics from a circular cylinder of diameter D are investigated experimentally in a free surface water channel. The vortex shedding is modified via the placement of a control cylinder of diameter \textit{D}/8 in the vicinity of the main cylinder. A methodology is presented to link changes in the wake dynamics and loading on the main cylinder. The analysis combines Fourier Modal Decomposition, Proper Orthogonal Decomposition, and phase averaging. Based on differences in the wake dynamics, the influence of the control cylinder can be classified according to its placement: (i) in the free stream outside the main cylinder shear layer; (ii) within the main cylinder shear layer; and (iii) in the recirculation region. While fluctuating lift is significantly reduced in all cases, the mean and fluctuating drag are affected differently. A generalized model-free method to optimize parameters for open-loop and closed-loop control in fluid mechanics applications is then presented. A multi-objective evolutionary algorithm (MOEA) is employed to minimize the oscillating lift caused by vortex shedding from the main cylinder. The control cylinder is prescribed a position as well as a periodic motion in two dimensions. The MOEA efficiently handles the larger optimization parameter space. The first objective of the algorithm is to minimize the fluctuating force coefficient $C_{L_{RMS}}$, while the second objective is to minimize of the actuation power required to drive the control cylinder. The final solution suppresses $C_{L_{RMS}}$ by over 90\% using near-zero actuation power. Further, the MOEA automatically provides a sensitivity study as to the influence of the different parameters and also in which spatial area the greatest influence is expressed.
- ItemOpen AccessMulti-sensory integration for intuitive interaction within virtual environments(2007) Widmer, Antoine; Hu, Yaoping
- ItemOpen AccessMultiuser Usability of Collaborative Virtual Environments(2017) Erfanian, Aida; Hu, Yaoping; Far, Behrouz H.; Yanushkevich, Svetlana; Wang, Ruisheng; Latoschik, Marc E.Collaborative virtual environments (VEs) require suitable interaction models for resolving conflicts and promoting multiuser usability. An interaction model is a key component of a collaborative VE. Traditional models such as the first-come-first-serve (FCFS) model have a problem of disregarding the vital socio-human need of equality in interaction (i.e., EII). This problem may impair the suitability of a model. Other components of a collaborative VE, including interaction devices and communication cues, may also affect the suitability of a model. Common cues are verbal and vibrotactile cues. Traditional usability studies on collaborative VEs suffer from several shortcomings. First, a set of multiuser usability metrics are not defined to consider socio-human needs and cover all possible usability factors presented by recent international standards. Secondly, suitable models to address these needs have not been sufficiently investigated. Finally, there have been a lack of studies that investigate the role of devices and cues on the suitability of models. To address these shortcomings, this thesis proposes a framework of multiuser usability for assessing collaborative VEs. The proposed framework consolidates socio-human needs and standard factors of usability. Moreover, a dynamic priority (DP) model that considers the vital need of EII is proposed to address the shortcomings of traditional models. The proposed DP model grants interaction opportunities to users based on the recency of their gained accesses. Investigations under the proposed framework indicated that compared to the FCFS model, the DP model yields perceived EII independent of devices and significantly improves the multiuser usability. The DP model also yields perceived EII regardless of cues. However, a combination of verbal and vibrotactile cues significantly promotes the multiuser usability of a VE governed by the DP model. These results imply the suitability of the DP model as well as combined verbal and vibrotactile cues to promote the multiuser usability within VEs.
- ItemOpen AccessRGB Predicted Depth Simultaneous Localization and Mapping (SLAM) for Outdoor Environment(2024-04-18) Brahmanage, Gayan Sampath; Leung, Henry; Wang, Yingxu; Hu, Yaoping; Bisheban, Mahdis; Gu, JasonThis thesis focuses on visual simultaneous localization and mapping (V-SLAM) for outdoor applications such as autonomous driving. While most V-SLAM methods have been tested on small-scale settings such as mobile robots, applying them in expansive outdoor spaces introduces additional complexities. The larger scale of the environment, dynamic obstacles, and depth-perception limitations of visual sensors pose challenges for V-SLAM methods. The first contribution introduces a dynamic V-SLAM approach. A novel front-end motion tracking approach is developed to recover multiple motions from image frames, considering key-points observed after map initialization as dynamic with time-varying locations. The proposed approach searches for key-point clusters based on their motion and classifies associated motions probabilistically. A bundle adjustment (BA) optimizes the local map, camera trajectory, and key-points motion within a unified V-SLAM system. BA maintains the geometric relationships between dynamic key-points and camera poses in the co-visibility graph, enhancing the overall robustness and accuracy of V-SLAM in populated environments. The second contribution of this thesis centers around a deep-learning-based depth prediction approach, which proves effective for estimating metric scale maps using a monocular camera. An unsupervised depth prediction approach is proposed using a novel convolution vision transformer (CViT) model architecture to infer depth from monocular images. The proposed encoder features a dual CViT block (DCViT); one block generates self-attention solely based on the spatial context of input feature vectors, and the other learns to generate attention based on the scene’s geometry. Contrastive learning of visual representations is applied to DCViT, where the model takes depth predictions from the same model through a feedback path as a supervisory signal to train the DCViT. Integration with residual blocks enables the learning of local and global receptive fields that produce predicted disparity maps at a higher level of detail and accuracy. Experimental results demonstrate significant improvements over state-of-the-art methods across multiple depth datasets. The third contribution of this thesis involves a comprehensive investigation into the utilization of predicted depth within monocular SLAM. This exploration aims to enhance the accuracy of map estimation in metric scale. Most existing approaches struggle with the non-Gaussian distribution inherent in heavy-tail noise produced by depth prediction models. The proposed monocular SLAM approach utilizes t-distribution for ego-motion, with parameter estimation achieved through maximum likelihood (ML) estimation using the expectation maximization (EM) algorithm. Experiments on real data show that the proposed t-distribution renders the monocular SLAM algorithm inherently robust to outliers and heavy-tail noise produced by depth prediction models.