Robotics · Computer vision · Sensor fusion · Autonomous systems
I work on robotic perception and intelligent systems, with a background in mechatronics and marine robotics. My recent research focuses on helping underwater robots acquire, enhance, interpret, and combine visual and sonar information. I am particularly interested in reliable perception under difficult imaging conditions, 3D object localisation, visual SLAM, and the connection between perception and robot control.
I completed my Erasmus Mundus MIR studies at Université de Toulon, France, and the Norwegian University of Science and Technology (NTNU), Norway, and submitted my master's thesis in July 2026. The final joint degree award is pending. I also hold a B.Sc. in Mechatronics Engineering from RUET, where I ranked 2nd in my class with a CGPA of 3.69/4.00.
| Project | Focus |
|---|---|
| Multi camera perception and fusion | ROS 2 detection, class aware association, confidence fusion, stereo calibration, and triangulation. |
| BlueROV sensor acquisition | Camera and sonar acquisition, MAVROS integration, recording, and operator tools for underwater experiments. |
| LoFTR stereo triangulation | Feature correspondences, stereo object association, and object level 3D localisation, with offline and ROS 2 workflows. |
| ORB SLAM3 with OceanSim | Integration notes, camera configuration, and launch support for monocular SLAM with ROS 2 and Isaac Sim. |
| AUV depth control with LQR | MATLAB and Simulink coursework on vehicle modelling, linearisation, and optimal depth control. |
| Visual servoing | Collaborative coursework on BlueROV visual tracking and image based control, including a Panda arm simulation. |
My thesis, “Deep Learning Algorithms for Multi-Camera Fusion of Information Acquired with Different Cameras Integrated on a BlueROV: Application to the Vision Enhancement and Detection of Unidentified Floating Objects,” brings together sensor acquisition, underwater image enhancement, object detection, and stereo localisation. The work includes synchronised camera and sonar acquisition, per camera detection and confidence fusion, stereo calibration, LoFTR correspondences, and experiments with visual SLAM and simulation.
I am preparing manuscripts on lightweight underwater image enhancement, a ROS 2 perception and sensor fusion framework for BlueROV, and a unified underwater object detection benchmark. The enhancement study includes TinyCurveNet CCM, AquaFastNet, and EdgeOSA. These manuscripts are in preparation; the repositories above document the publicly available parts of the work.
| Area | Repositories |
|---|---|
| Underwater detection | YOLOv8 experiments |
| Vehicle modelling | Sparus AUV simulation |
| State estimation | Kalman filtering notebooks |
| Applied machine learning | Electric vehicle range prediction |
| Data preparation | NASA battery data conversion |
| Scientific software | Rust terrain viewer |
Programming and tools: Python, C++, MATLAB, Simulink, Rust, Git, and Linux.
Perception and learning: PyTorch, OpenCV, YOLO, stereo vision, image enhancement, and feature matching.
Robotics: ROS 2, BlueROV, MAVROS, visual servoing, state estimation, and AUV modelling and control.
Simulation: NVIDIA Isaac Sim, OceanSim, and PyBullet.
My publication record includes eight journal articles and two conference papers, alongside the manuscripts currently in preparation. I am open to PhD, research, and engineering opportunities in robotics, computer vision, intelligent perception, and autonomous systems.
Profile information updated in September 2026.