DR-NTU (Digital Repository of NTU)
Not a member yet
116018 research outputs found
Sort by
In-Digestion
No full text“In-Digestion” is a 2D animated fantasy film about a pair of elf siblings adventuring in the unknown. The film follows older sister Penny and younger brother Odi as they explore an unfamiliar forest and meet some weird creatures. The story theme is a conflicting sibling dynamic thrown into an inner world exploration genre. Despite being a simple story premise, various story elements were used to subvert exceptions. These helped to enhance the narrative, creating deeper emotions and memorability for the viewers.Bachelor's degre
Multi-modal driver action recognition
No full textWith the rise of self-driving vehicles, the issue of driver action recognition has never been more
pertinent. However, current Driver Monitoring Systems (DMSs) utilize only a single modality,
which leaves these systems susceptible to the challenging conditions of real-world driving.
Through this report, we propose the development of a multimodal driver action recognition
model, leveraging RGB, InfraRed and Depth sensor inputs to create a more robust model
capable of performing under varying conditions.
We implement this model by improving upon the Multi-Modal Video Transformer (MM-ViT)
architecture, a Transformer-based model for video classification tasks using a pretrained ViT
backbone. We modify this model by introducing a modified Multi-Headed-Relational Attention
(MHRA) model to explicitly model and distinguish between local and global features in the
spatial and temporal dimensions. We also implement a computationally efficient Shift-Merge
Attention mechanism to handle modality fusion, which shifts feature channels in the data to
capture inter-modal relationships at zero added computational cost.
Our model outperforms other State-Of-The-Art Convolutional Neural Network (CNN)- based
models and achieves comparable performance to other transformer-based models when trained
on the Drive and Act dataset, a multimodal dataset for driver action recognition.Bachelor's degre
Moving discretized control set model predictive control for dual active bridge-based battery charger
No full textThis paper presents an improved moving discretized control set model predictive control (MDCS-MPC) scheme for a dual active bridge (DAB) converter-based battery charger, aimed at achieving improved current regulation performance with reduced computational demand. To achieve these objectives, the proposed strategy incorporates a reduced order average model of the DAB for precise state prediction and utilizes a hybrid PI-MPC framework that selectively excludes half of the control set candidates. Simulation results validate the proposed scheme’s superior dynamic response and steady-state performance, confirming its suitability for advanced battery charging applications.Agency for Science, Technology and Research (A*STAR)Submitted/Accepted versionThis research is supported by A*STAR under its RIE2025 Manufacturing Trade and Connectivity (MTC) Industry Alignment Fund - Pre- Positioning (IAF-PP), with Award No. M23L5a0002
Development of MATLAB code to apply direct stiffness method for converting a finite element model to a "boundary node" model
No full textThe Finite Element Method (FEM) is a widely used numerical technique in mechanical, civil, and structural engineering for applications such as stress analysis, vibration response, and static displacement evaluation. While commercial software like ANSYS offers comprehensive FEM capabilities, developing an in-house code provides greater control and is particularly valuable for research work.
This project focuses on developing a finite element code in MATLAB, based on the Direct Stiffness Method (DSM). The aim is to construct a reduced “boundary node” model (keeping only the boundary nodes of the finite element model) capable of producing results comparable to that of the original finite element model. The results of the original model are obtained from ANSYS Mechanical APDL. Stiffness matrices are first generated in ANSYS and then extracted for use in MATLAB. The MATLAB code reads the stiffness matrices, assembles the global stiffness matrix using nodal connectivity, and solves the system to obtain nodal displacements.
Three sets of validation examples are evaluated under twelve displacement scenarios grouped into three exercises. Displacement results from MATLAB code are compared with ANSYS outputs, with a focus on displacement errors. The results match closely for all loading condition tested, with observed errors typically below 1%. However, in cases involving complex or multi-directional loading, higher errors (2%) were observed.
The overall objective of this project is to reduce the number of nodes in the model by keeping only the boundary nodes, which will lead to a decrease in computational time. The percentage decrease in the number of nodes for the final test problem is 25%.Bachelor's degre
Multi-camera calibration using prior map and dynamics for mini-UAV system
No full textThis report involves developing a camera calibration system for Unmanned Aerial
Vehicle (UAV) that combines mapping information and motion dynamics assessment
to improve accuracy of calibration. UAV systems with dynamic operations require a
new calibration method that uses spatial data together with motion pattern analysis.
The research creates a framework which uses adaptive calibration methods to
enhance camera alignment while establishing greater image accuracy for UAV-based
monitoring systems. The experimental results reflect better calibration precision over
traditional methods of calibration, therefore proving practical use for aerial
monitoring, search and rescue operations, infrastructure assessments and many more.
Additional future developments should normalise the approach to operate under a
variety of environmental conditions while utilising multiple UAV types to achieve
maximum adaptability.Bachelor's degre
Leveling up maritime education: the role of gamification in maritime learning and industry readiness
No full textIn the field of maritime education, where motivation, engagement, and readiness for the job
market are essential, gamification has emerged as a transformative educational tool. This
study explores the impact of gamified learning environments on knowledge acquisition, skill
development, intrinsic motivation, student engagement, and preparedness for the industry.
To assess students' experiences with gamified learning platforms, structured surveys, a game
we developed (“Oh Ship!”) and a quantitative research methodology was utilised. The study
applied factor analysis to analyse the six constructs, namely, “Gamification Elements”,
“Student Engagement”, “Intrinsic Motivation”, “Knowledge Acquisition”, “Skill
Development”, and “Industry Readiness”. Reliability tests were conducted to evaluate the
credibility and consistency of the data. The relationship between the constructs was
subsequently examined using multiple regression model analysis. The results indicate that
gamification elements significantly enhance student engagement, thereby fostering intrinsic
motivation and improving learning outcomes. Additionally, the study revealed that both
knowledge acquisition and skill development play a critical role in shaping industry
preparedness, highlighting the potential of gamification to equip students for real-world
challenges in the maritime sector.
The study further examines the potential of these technologies to enhance the effectiveness of
learning through gaming. Recommendations include utilising social media as an interactive
learning resource, integrating gamification into educational curricula, and employing
advanced simulation-based training methods. By demonstrating how gamification can bridge
the gap between conventional education and industry standards, this study contributes to the
growing domain of technology-enhanced learning and equips students with the necessary
skills to thrive in the marine industry.Bachelor's degre
Control and simulation of an aerial robot tracking a ground robot
No full textThis project presents the simulation and control of a cooperative robotic system, where a 6DOF quadcopter autonomously follows a differential-drive ground robot. Both robots are modelled using MATLAB Simulink with their respective dynamics and controlled by a modular control framework. The quadcopter uses a cascaded control structure that separates position and attitude, while the ground robot is guided by nonlinear controllers for trajectory and orientation tracking. A second-order tracking differentiators is used to act as a smoother to enhance signal and control accuracy. The result of the Simulation demonstrates demonstrate precise tracking performance for both the quadcopter and ground robot. This result will be shown in the later portion of the report where one will see the close alignment between the desired and actual position along the x, y, and z axes.Bachelor's degre
Application of the five factor model of personality with impulsivity as a mediating factor on electronic cigarette attitudes in Singapore
No full textElectronic cigarette (e-cigarette) use is an emerging phenomenon with severe health
outcomes. This study examined attitudes on e-cigarette use and how the Big Five personality
traits and impulsivity correlated with said attitudes. The mediating role of impulsivity on the
personality-e-cigarette attitude relationship was investigated. A cross-sectional study using a
survey (N=178) collected data on e-cigarette attitudes, impulsivity, and personality traits.
Descriptive statistics, Spearman’s rank correlations and bootstrapping analyses suggested
conservative views on e-cigarettes use in Singapore, with majority of participants perceiving
high risks of e-cigarette use (>80% acknowledgement). Results on perceived benefits and
policy stances were more mixed. Conscientiousness, rs(176) = -.11, p = .080, and Openness
to Experience, rs(176) = -.12, p = .055, were found to correlate significantly with more
permissive policy stances, Extraversion was reported to negatively correlate significantly
with less permissive attitudes, rs(176) = .13, p = .040, and less perceived benefits, rs(176)
= .11, p = .073. Impulsivity, rs(176) = .14, p = .034, was also found to significantly correlate
positively with less permissive attitudes. While weak, there was also some evidence
suggesting possible mediation roles of impulsivity in the personality-e-cigarette attitude
relationship. Findings may help to inform policy and public education efforts to discourage e-cigarette use in Singapore.Bachelor's degre
3D concrete printing
No full textThe construction industry continues to lag behind other sectors in productivity, highlighting the
urgent need for innovative approaches. A promising solution is extrusion-based 3D Concrete Printing (3DCP). When combined with the unique properties exhibited by Engineered Cementitious
Composites (ECC), 3DP-ECC has the potential to address both economical and sustainability
challenges. However, achieving a balance between the conflicting demands of pumpability and
buildability remains a significant hurdle to this process. Therefore, this study aims to develop a lowcarbon ECC capable of 3D printing by incorporating granulated blast furnace slag (GGBS) and
rejected brine, while also enhancing its self-healing capabilities. Results saw the successful extrusion of a printed element with satisfactory print qualities. However, achieving sufficient tensile strain proved to be a difficult task. Contrary to trends reported in past literature, cast specimens in this study also exhibited better tensile performance compared to printed samples. Furthermore, mixes containing GGBS and rejected brine demonstrated slower self-healing compared to pure OPC based mixes over five and seven wet/dry cycles, possibly due to their low rate of hydration. Given these findings, future research should explore strategies to better balance mechanical performance and printability and implement more comprehensive methods to assess the self-healing behaviour of 3DPECC.Bachelor's degre
OptRCA: a more efficient and accurate approach for automated root cause analysis and explanation
No full textWith the development of automated software testing technology, software developers can get a large number of crash test cases in a short period of time. However, analyzing these crash test cases and finding their root cause is a time-consuming and labor-intensive task. Techniques based on reverse execution and backward taint analysis are proposed to locate the root cause, but can’t provide context information or explanation of the underlying fault. To address these two limitations, researchers have proposed an automated root cause analysis technique called AURORA. Although this technique provides powerful root cause analysis capabilities, it also have two obvious shortcomings. First, the results of root cause analysis are not accurate enough. Second, the efficiency of root cause analysis is not high enough. In order to improve these two shortcomings, we propose OptRCA, a more efficient and accurate approach for root cause analysis and explanation. Like AURORA's fuzzing strategy, OptRCA is also designed based on AFL's crash mode. The difference between them is mainly reflected in three points. First of all, the goal pursued by OptRCA is different from that of normal fuzzing technology. OptRCA pursues maximum correlation to ensure that as many crash test cases as possible are related to the same root cause. This test case with maximum correlation can greatly improve the accuracy of root cause analysis. Second, OptRCA proposed a more efficient non-crash test case retention strategy, which we named “Hill Climbing Retention”. Using the hill climbing retention method, OptRCA can obtain sufficient root cause information while retaining only a few non-crash test cases. Since the number of test cases is greatly reduced, the efficiency of OptRCA's subsequent root cause analysis process is also greatly improved. In addition, OptRCA also optimizes the analysis formula to obtain more accurate analysis results. In the evaluation experimental results, OptRCA is significantly better than AURORA in terms of accuracy and efficiency. Quantitative analysis shows that OptRCA is 65% more accurate and 61% more efficient than AURORA.Submitted/Accepted versio