About the Project
Video and images are now common elements of multimodal sensor packages on terrestrial vehicles, like self-driving cars and aerial drones, of all shapes and sizes. Yet these modalities have seen limited use for adaptive sampling approaches for Autonomous Underwater Vehicles (AUVs). Cameras are typically deployed underwater purely as a data collection tool, programmed to capture visual data at a set rate for further analysis on the surface.
Using cameras on AUVs to inform real-time behaviour requires creative computer vision (CV) approaches robust to the challenges of operating in the underexplored deep sea; models are severely power limited, the image quality is highly variable, environments are sometimes visually dense, and the sampling distribution is typically open-set and long-tailed. We will explore novel CV approaches to estimate biological populations and compute statistically grounded uncertainty bounds to inform decision making while underway. The project touches on diverse subjects in AI and CV, including handling distribution shift, domain generalization, few- or zero-shot learning, active learning, and geospatial learning approaches.
Eligibility criteria
Applicants are expected to have a strong background in AI, computer vision, and/or robotics. Previous work in ecology or ocean science is desirable but not required. A keen interest in those topics is a necessity.
King's expects applicants to hold a First-Class Honours degree at undergraduate level or a Distinction at Master's level (or their international equivalents). Applicants who do not meet these criteria may be able to demonstrate their competence through alternative evidence of achievement or experience.
Application Process
Applicants are encouraged to contact Dr Eric Orenstein (eric.orenstein@kcl.ac.uk) to discuss the position informally before applying. Once you have spoken with us and all are happy to proceed, please follow the formal application process below.
Formal Application Process
To be considered for the position, candidates must apply through the King's Apply online application system. Full details are available on the Research degrees | Department of Informatics | King's College London
When applying, please select:
- Programme: Computer Science Research MPhil/PhD (Full-time)
- Supervisor: Eric Orenstein
- Project title:
- In the Funding section of the form:
- Select Option 5 – "I am applying for a funding award or scholarship administered by King's College London."
- Enter the following code exactly as written in the 'Award Scheme Code or Name' box: 1138
The selection process will involve a pre-selection on documents and, if selected, will be followed by an invitation to an interview. If successful at the interview, an offer will be provided in due course.
References
Katija et al. 2021 "Visual tracking of deepwater animals using machine learning-controlled robotic underwater vehicles." WACV.
Lange et al. 2024 "Active learning-based species range estimation." NeurIPS.
Perez et al. 2024 "DISCount: Counting in Large Image Collections with Detector-Based Importance Sampling." AAAI.
Hughes et al. 2021 "Sampling biases shape our view of the natural world." Ecography.
Andersson et al. 2023 "Environmental sensor placement with convolutional Gaussian neural processes." EDS.
Yao et al. 2022 "Wild-Time: A Benchmark of in-the-Wild Distribution Shift over Time." NeurIPS.
Pollock et al. 2025 "Harnessing artificial intelligence to fill global shortfalls in biodiversity knowledge." Nature Revs.
Funding Notes
A fully-funded studentship for 3.5 years starting in October 2027.
Stipend: £24,514 per annum, subject to increase
Bench Fees: £1,500 per annum
Tuition fees: covered in full for home students
