Kulunu Samarawickrama

I am Kulunu Samarawickrama, a doctoral researcher and robotics software developer working at the intersection of robotic manipulation, 3D perception, and autonomous systems. My work focuses on building perception pipelines that help robots understand cluttered scenes, localize objects, and act reliably in real-world environments.

My background combines computer vision, machine learning, and practical system integration. I am especially interested in how RGB-D sensing, point cloud methods, simulation, and learning-based models can be brought together into end-to-end systems for manipulation and assembly. Alongside research, I care about writing clean, reusable code and building workflows that make experiments easier to reproduce and extend.

Across my projects, I have worked on ROS2-based perception stacks, synthetic data generation, simulation-to-real evaluation, and benchmarking pipelines for robotics. I enjoy turning research ideas into working software, whether that means prototyping perception models, connecting them to robot interfaces, or validating them at scale in simulation and HPC environments.

My current interests include:

• robotic manipulation and autonomous systems
• 3D perception, RGB-D sensing, and point cloud understanding
• simulation-to-real transfer and synthetic data pipelines
• reproducible research software and scalable experimental workflows

This site brings together my publications, projects, software, and notes. It is a place to document the work I am doing, share what I am learning, and connect research ideas with practical engineering. Feel free to explore and get in touch if our interests overlap.