Norsk English

Object Localization and pose estimation for remotely operated vehicles (ROVs)

In this thesis you will be given access to real underwater images/video along with CAD models for selected objects and evaluate and demonstrate 6D pose estimation algorithms.

Contact person

Ill.: SINTEF

Determining the 3D translation and rotation of objects is fundamental for enabling autonomous grasping with uncrewed underwater vehicles such as remotely operated vehicles (ROVs). Having access to data is essential for progressing research in underwater robotic systems, facilitating method evaluation, and gaining a better understanding of their constraints.

Objective

  • Establish an underwater 6D pose dataset for diverse underwater industrial objects on data from the field
  • Develop and enhance image based 6D pose estimation and object detection algorithms for underwater based on classical computer vision and deep learning methods.

Tasks

  • Develop code for dataset creation and benchmarking procedures using markers.
  • Design a set of benchmarks to evaluate ground truth pose with respect to object CAD models
  • Implement and test a baseline using a state-of-the-art 6D pose estimation algorithm.
  • Improve and optimize for underwater water real-time demonstration
  • Prepare documentation and guidelines for dataset usage and benchmark evaluation.

Qualifications

Current Masters degree student in cybernetics or computer science, hands on experience on Computer vision, good Python programming skills, deep learning, and object detection

Expected Results and Learning Outcome

The results of the thesis will be the methodology and the implementation state of the art 6D pose estimation techniques. The best performing technique would be demonstrated in the lab or precaptured underwater videos.

  • Supervisor:  Ahmed Mohammed
  • Internship period: Jan 2025 – June 2025
  • Location: Oslo, Norway
  • Project: Safe and autonomous subsea intervention (Safesub)

References

  1. Mohammed, A., Kvam, J., T. Thielemann, J., Haugholt, K. H., & Risholm, P. (2021). 6D pose estimation for subsea intervention in turbid waters. Electronics, 10(19), 2369.
  2. Risholm, P., Ivarsen, P. Ø., Haugholt, K. H., & Mohammed, A. (2021). Underwater marker-based pose-estimation with associated uncertainty. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 3713-3721).
  3. Transeth, Aksel A., Ingrid Schjølberg, Anastasios M. Lekkas, Petter Risholm, Ahmed Mohammed, Martin Skaldebø, Bent OA Haugaløkken, Magnus Bjerkeng, Maria Tsiourva, and Frederic Py. "Autonomous subsea intervention (SEAVENTION)." IFAC-PapersOnLine 55, no. 31 (2022): 387-394.