Sagar Dalai

Mr. Sagar Dalai

orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 February 2025 in Machines
Verified
Journal Article
Open access
PDF file
MAHAMMAD IRFANMAHAMMAD IRFAN
MAHAMMAD IRFAN
SDSagar Dalai
Sagar Dalai
PTPetar Trslic
Petar Trslic
JRJames Riordan
James Riordan
GDGerard Dooly
Gerard Dooly
Unmanned aerial vehicle (UAV) state estimation is fundamental across applications like robot navigation, autonomous driving, virtual reality (VR), and augmented reality (AR). This research highlights the critical role of robust state estimation in ensuring safe and efficient autonomous UAV navigation, particularly in challenging environments. We propose a deep learning-based adaptive sensor fusion framework for UAV state estimation, integrating multi-sensor data from stereo cameras, an IMU, two 3D LiDAR’s, and GPS. The framework dynamically adjusts fusion weights in real time using a long short-term memory (LSTM) model, enhancing robustness under diverse conditions such as illumination changes, structureless environments, degraded GPS signals, or complete signal loss where traditional single-sensor SLAM methods often fail. Validated on an in-house integrated UAV platform and evaluated against high-precision RTK ground truth, the algorithm incorporates deep learning-predicted fusion weights into an optimization-based odometry pipeline. The system delivers robust, consistent, and accurate state estimation, outperforming state-of-the-art techniques. Experimental results demonstrate its adaptability and effectiveness across challenging scenarios, showcasing significant advancements in UAV autonomy and reliability through the synergistic integration of deep learning and sensor fusion.

27 September 2024 in IEEE Transactions on Intelligent Vehicles
Verified
Journal Article
Open access
MAHAMMAD IRFANMAHAMMAD IRFAN
MAHAMMAD IRFAN
SDSagar Dalai
Sagar Dalai
PTPetar Trslic
Petar Trslic
MSMatheus Santos
Matheus Santos
JRJames Riordan
James Riordan
GDGerard Dooly
Gerard Dooly
With the development of Autonomous Unmanned Aerial Vehicle's (UAV's), Precise state estimation is a fundamental aspect of autonomous flight and plays a critical role in enabling robots specially in GPS denied environment to operate safely, reliably, and effectively across a wide range of applications and operational scenarios. In this paper, we propose a tightly-coupled multi-sensor filtering framework for robust UAV/UGV state estimation, which integrates data from an Inertial Measurement Unit (IMU), a stereo camera, GPS, and 3D range measurements from two Light Detection and Ranging (LiDAR) sensors. The proposed LGVINS system significantly improves the accuracy and robustness of state estimation in both structured and unstructured outdoor environments, such as bridge inspections, open fields, urban city and areas near buildings. It also improves positioning accuracy in scenarios with or without GPS signals. The goal is to exploit the fact that these sensor modalities have mutually exclusive strengths, the visual, inertial and the Lidar sensor techniques are implemented to compensate for the robots state estimate errors in multiple outdoor challenging environment. It effectively reduces long-term trajectory drift and ensures smooth, continuous state estimation, regardless of GPS satellite availability. We demonstrate and evaluate the LGVINS approach on public dataset as well as our own dataset collected from the proposed hardware integration on UAV, deployed on computationally-constrained systems. This demonstrates that the proposed system achieves higher accuracy and robustness in state estimation across various environments compared to currently available methods.

6 June 2025 in Remote Sensing
Verified
Journal Article
Open access
PDF file
MMMarco Moreno
Marco Moreno
SDSagar Dalai
Sagar Dalai
GCGrace Cott
Grace Cott
BBBen Bartlett
Ben Bartlett
MSMatheus Santos
Matheus Santos
GDGerard Dooly
Gerard Dooly
Accurate classification of salt marsh vegetation is vital for conservation efforts and environmental monitoring, particularly given the critical role these ecosystems play as carbon sinks. Understanding and quantifying the extent and types of habitats present in Ireland is essential to support national biodiversity goals and climate action plans. Unmanned Aerial Vehicles (UAVs) equipped with optical sensors offer a powerful means of mapping vegetation in these areas. However, many current studies rely on single-sensor approaches, which can constrain the accuracy of classification and limit our understanding of complex habitat dynamics. This study evaluates the integration of Red-Green-Blue (RGB), Multispectral Imaging (MSI), and Hyperspectral Imaging (HSI) to improve species classification compared to using individual sensors. UAV surveys were conducted with RGB, MSI, and HSI sensors, and the collected data were classified using Random Forest (RF), Spectral Angle Mapper (SAM), and Support Vector Machine (SVM) algorithms. The classification performance was assessed using Overall Accuracy (OA), Kappa Coefficient (k), Producer’s Accuracy (PA), and User’s Accuracy (UA), for both individual sensor datasets and the fused dataset generated via band stacking. The multi-camera approach achieved a 97% classification accuracy, surpassing the highest accuracy obtained by a single sensor (HSI, 92%). This demonstrates that data fusion and band reduction techniques improve species differentiation, particularly for vegetation with overlapping spectral signatures. The results suggest that multi-sensor UAV systems offer a cost-effective and efficient approach to ecosystem monitoring, biodiversity assessment, and conservation planning.

27 April 2026 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
SDSagar Dalai
Sagar Dalai
MMMarco Moreno
Marco Moreno
JRJames Riordan
James Riordan
Thomas NeweThomas Newe
Thomas Newe
EOEoin O'Connell
Eoin O'Connell
GDGerard Dooly
Gerard Dooly
Hyperspectral image (HSI) classification is vital for environmental monitoring, land cover mapping, and precision agriculture, but its effectiveness is often constrained by the scarcity of labeled samples and high spectral similarity among classes. To address these challenges, we propose CrossCapsViT, a hybrid classification framework that integrates Capsule Networks (CapsNets) and Vision Transformers (ViTs) through a Cross-Attention Fusion (CAF) mechanism and a cross-layer adaptive fusion module, enabling richer and more discriminative spectral–spatial feature learning. To further improve efficiency in data scarce scenarios, we embed an Actor–Critic reinforcement learning based active learning (RAL) strategy that jointly leverages accuracy, uncertainty, and diversity in the reward structure, guiding the selection of the most informative samples while reducing labeling effort. Experiments conducted on four benchmark datasets (KSC, PU, HU2013, and Salinas) and a custom UAV based saltmarsh dataset (Derrymore, collected with a Pika-L sensor) demonstrate that CrossCapsViT with RAL consistently outperforms CapsViT and other baseline models in terms of classification accuracy, robustness, and generalizability. The proposed framework achieves up to 25% improvement in class-level accuracy on challenging vegetation classes, while reducing dependence on large annotated datasets, highlighting its potential for practical deployment in real-world ecological monitoring and remote sensing applications.

7 August 2026 in Remote Sensing
Verified
Journal Article
Open access
PDF file
SDSagar Dalai
Sagar Dalai
MMMarco Moreno
Marco Moreno
EOEoin O'Connell
Eoin O'Connell
Thomas NeweThomas Newe
Thomas Newe
GDGerard Dooly
Gerard Dooly
Hyperspectral image (HSI) classification is central to environmental monitoring, yet real-time deployment of deep learning models on resource-constrained edge platforms remains challenging due to high spectral dimensionality and computational overhead. In this paper, we propose TinyCapsViT, an ultra-lightweight hybrid architecture that integrates convolutional feature extraction, transformer-based attention, and capsule-inspired representation learning for efficient HSI classification under strict TinyML constraints. The model employs a minimal convolutional stem using pointwise and depthwise separable convolutions to capture local spectral-spatial features, followed by a compact tokenization strategy and learnable positional embeddings. A lightweight self-attention module enables global context modeling with reduced computational complexity, while a capsule-inspired refinement block with squash nonlinearity and residual scaling enhances feature discrimination. The proposed architecture contains only 2781 trainable parameters, representing up to a 30× reduction compared with larger architectures such as ResNet and Vision Transformer, and approximately 11× and 9.5× fewer parameters than the CNN and 3D-CNN baselines, respectively. Extensive experiments across benchmark hyperspectral datasets demonstrate that TinyCapsViT achieves up to 99.61% overall accuracy and maintains competitive classification performance despite its substantially reduced model complexity. Although TinyCapsViT does not consistently achieve the highest classification accuracy compared with larger baseline models, it provides a favourable trade-off between classification performance and computational efficiency. These results demonstrate the potential of TinyCapsViT as a practical solution for real-time hyperspectral analysis on resource-constrained edge platforms and UAV-based environmental monitoring applications.

19 August 2026 in Mathematics
Verified
Journal Article
Open access
PDF file
JAJose Alvarez
Jose Alvarez
MMMarco Moreno
Marco Moreno
SDSagar Dalai
Sagar Dalai
MSMatheus Santos
Matheus Santos
GDGerard Dooly
Gerard Dooly
The study of galactic dynamics remains fragmented across three competing paradigms—Dark Matter (ΛCDM), Modified Newtonian Dynamics (MOND), and Yukawa-type modifications. This work relies on Enhanced Newtonian Dynamics (END), a parameter-free gravitational framework that derives rotational velocities exclusively from observable baryonic mass distributions and geometric configurations. Applying END to 39 high-quality galaxies from the SPARC database via a three-oblate spheroid composite model (core, bulge, and disc), we achieve mean velocity errors below 3% for 71.79% of the sample, with a median χ2=1.77. The fundamental END relation, MT2=κV (where κ≈1.4121×1011kgm3s2), successfully reproduces observed rotation curves without invoking non-baryonic dark matter haloes or modified gravity interpolation functions. Notable exceptions (e.g., IC2574, χ2=51.56) arise from geometric mismatches between the assumed exponential decay profiles and the actual density structures of dwarf irregular systems. Classical Newtonian Dynamics, applied to identical mass distributions, accounts for only 20% of observed velocities, suggesting the “missing mass” problem reflects limitations in how gravitational law relates enclosed mass to orbital motion rather than unseen matter. END’s zero-parameter architecture renders it strictly falsifiable, offering a parsimonious alternative to parameter-heavy competing models while requiring future iterations to incorporate adaptive geometric baselines for morphologically diverse galaxies.