Autonomous maritime navigation is set to transform the maritime industry providing consistent, continuous and cost-effective navigation free from human error. However, there remain a number of barriers to its full adoption including trust and regulation acceptance. One such regulation-based restraint is the International Maritime Organisation's (IMO) Collision Regulations (COLREGs), written to ensure safe navigation through open water. These COLREGs require a "look-out", a slightly ambiguous term that could be used to describe a human or equivalent automated system. Currently, this function is performed on almost all vessels by a human watchkeeper, but in the future could this be replaced with a computer?
Rule 5 of the COLREGs specifically requires that "every vessel shall at all times maintain a proper look-out by sight and hearing as well as by all available means appropriate in the prevailing circumstances and conditions so as to make a full appraisal of the situation and of the risk of collision.”
For a computerised system to take on this “look-out” role it must incorporate robust computer vision & AI in all weathers and light conditions. It is expected that such a system could compare the input data against Radar and Automatic Identification System (AIS) data to build up a complete picture of both the near field and distant vessels and hazards.
Object detection and vision-based ship navigation is an essential task for autonomous ship navigation; however sunlight reflection, camera motion and illumination changes may cause false object detection in the maritime environment. Deep learning based algorithms are already developed in open literature [Zhang et al. (2021)] but are yet to provide a trustworthy all weather/light solution.
The proposed project will focus on creating a computer vision system which detects and classifies ships in different light conditions and weather conditions specifically focusing on interpreting night-time navigation lights and day shapes in transitional light periods as dawn and dusk. This computer vision output will then need to be processed, to interpret the pattern of lights and shapes to determine the orientation and status of the oncoming vessel. This capability will be especially useful in harbour approaches and congested waterways where room for manoeuvrability is limited and the look-out role is critical. In such a scenario it may be that the AI system could run in parallel with a human look-out to build trust and provide an alert system to prevent human error. Currently, deep neural networks are utilised with good effect in ship detection and classification using high resolution images. However often in real life, and at sea, imagery can be poor quality due to difficult weather conditions and low-resolution cameras. This project will aim to develop solutions for the detection and classification in all-weather and light conditions such as at dawn and at dusk. In order for the system to be trusted to operate effectively, it will need to provide an equivalent performance to a human viewing the image. The programme will need to be able to capture if a vessel is coming towards them, away from them, and if they have any additional navigational constraints or navigational shapes.
This project is supported by BMT - a maritime-orientated high-end design house and technical consulting firm driven by a passion for solving complex, real-world problems.
With around 1,500 professionals located in 47 offices in the Americas, Asia, Australia, and Europe, BMT draws upon a wide range of experience and expertise to provide high-quality, high-value products and services. From initial concept through to design, construction, operation, and eventual decommissioning, BMT support customers at every stage of the project lifecycle.
As a external project supporter, BMT can provide real world context and input into the project, ensuring long term exploitation.
This project is associated with the UKRI Centre for Doctoral Training (CDT) in Accountable, Responsible and Transparent AI (ART-AI). We value people from different life experiences with a passion for research. The CDT's mission is to graduate diverse specialists with perspectives who can go out in the world and make a difference.
Applicants should hold, or expect to receive, a First or Upper Second Class Honours degree in a relevant subject. A master’s level qualification would be advantageous. A strong background in mathematics, good programming skill and experience of machine learning are highly desirable.
Formal applications should be made via the University of Bath’s online application form. Enquiries about the application process should be sent to [Email Address Removed].
Start date: 3 October 2022.