Overview
Oceanbird is developing rigid wing sails to reduce fuel consumption on commercial vessels by harnessing wind energy. The rigid wing sail is made with steel and composite parts, which move but also bend and deflect in foreseeable and unforeseeable ways. Accurately quantifying bending and twisting of moving structures is critical for structural integrity models, yet most vision-based deflection studies target static bridges or buildings. This project explores whether a hybrid machine-vision and AI approach can supplement traditional point-sensor techniques when the monitored component is a land-mounted rigid wing sail that rotates during testing. The study will benchmark state-of-the-art depth-estimation neural networks, stereo/monocular vision pipelines, and classic photogrammetry against high-accuracy reference sensors to determine achievable precision, repeatability, and long-term stability.
Objective
This thesis aims to develop a framework to perform controlled deflections/bending in composites and steel objects and to capture video with optimized camera/marker layout. The results should then be quantified against benchmarked state of the art depth-estimation neural networks and photogrammetry methods. If time permits, evolve the framework for dynamic systems.
Research Platform
There is possibility to conduct actual measurements on Oceanbird’s 40 m tall land-based prototype in Landskrona, Sweden. But the important deliverable would be a technical feasibility study conducted on either simulation or a scaled model.
Skills
We are looking for a driven student with strong research mindset, with ability to design experiments, analyze results, document and iterate solutions. It is beneficial to have strong software skills, especially machine vision, with a basic knowledge about structures or a strong will to learn the above. It is possible to tweak the proposal and objectives to align with your interests and your degree requirements.
Contact
Akash Singh, Control System Engineer, akash.singh@theoceanbird.com