Real-time wave imaging using stereo and monocular machine vision

Applications are invited for a self-funded, 3-year full-time or 6-year part time PhD project.

The PhD will be based in the School of Mechanical and Design Engineering and will be supervised by Dr Ya Huang, Dr Andrea Bucchi and Professor Hui Yu. 

The work on this project could:

• Compare and evaluate existing stereo-vision pipelines of surface wave 3D reconstruction using 1) epipolar semi-global search method generating a dense 3D map but with long processing time, 2) a fast real-time version using sparce feature matching methods to exploit frequency dispersion relation of gravity sea waves and Fast Fourier Transform to continuously update the 3D surface and 2D wavenumber spectrum estimations; 3) a fast real-time convolution neural network (CNN) architecture to fill the gaps of sparce 3D points with a hydrodynamic ‘dispersion’ model to effectively increase sample density. 
• Create a representative stereo vision wave imaging database on shore and on moving test boats and lifeboats. Apply superpixel tracking and image abstraction, and machine learning methods to further reduce processing time for wave semantic feature extraction.
• Incorporate a hydrodynamic wave model library to simulate and classify detailed wave-induced load on the vessel. Match estimation of wave hydrodynamic parameters between the observation and the wave models.
• Construct a monocular vision model using the stereo vision database in a CNN framework to further reduce time complexity for real-time execution.

Project description

The project intents to develop a real-time machine vision feature extraction pipeline of oncoming sea waves at close proximity. The new algorithm will make possible a real-time seakeeping, dynamic motion planning, and shock mitigation controller onboard both manned and unmanned vessels. The outcome will devise understanding and correlation between close-range wave characteristics with dynamic loads transmitted to the vessel structure and human occupants. This is part of a wider project to help marine search and rescue vessels to handle extreme sea conditions in a safer and more efficient scheme currently funded by the Royal National Lifeboat Institution (RNLI) investigating crew bracing strategies and next generation lifeboat design. 

The School of Mechanical and Design Engineering owns two small-scale unmanned surface test vessels, and has access to a University owned power boat with stereo vision camera system dedicated to this project for wave data collection and machine vision algorithm validation. The project will start with comparing and adapting existing stereo vision techniques to achieve off-line wave characterisation. The new in-the-loop feature detector and the unique close-wave imagery data made available will make a significant contribution to the wider machine vision community for outdoor mobile systems and the marine industry. 

The visual features will form an important part of the seakeeping and shock-mitigating navigational decisions that are key to the future surface vessels for safer operations. The project is aligned with the University’s vision to build global and national partnership through the boundary-breaking themes of future transportation and intelligent systems.  

The project offers great opportunities to engage with a wide spectrum of industrial collaborators from the fast marine craft suppliers, marine robotic companies, underwater inspection firms, and coastguard agencies. The student is expected to collaborate with project partners and attend conferences, project meetings, and workshops.

General admissions criteria

You'll need a good first degree from an internationally recognised university or a Master’s degree in an appropriate subject. In exceptional cases, we may consider equivalent professional experience and/or qualifications. English language proficiency at a minimum of IELTS band 6.5 with no component score below 6.0.

Specific candidate requirements

You should have good subject knowledge of engineering analytical skills, programming, mathematics and programming.

Knowledge of robotics, computer science, software engineering, electronics, mechanics, physics, computer graphics, machine vision and deep learning would be desirable. .   

How to Apply

We encourage you to contact Dr Ya Huang ([Email Address Removed]) to discuss your interest before you apply, quoting the project code below.

When you are ready to apply, please follow the 'Apply now' link on the Mechanical and Design Engineering PhD subject area page and select the link for the relevant intake. Make sure you submit a personal statement, proof of your degrees and grades, details of two referees, proof of your English language proficiency and an up-to-date CV. Our ‘How to Apply’ page offers further guidance on the PhD application process. 

When applying please quote project code:SMDE6101023