Today’s marine science requires the collection of data in extreme environments, including polar oceans and under ice. Autonomous Underwater Vehicles (AUVs) are often employed for these tasks, granting access to areas not reachable with conventional ship-based sampling methods. The availability of a reliable obstacle avoidance system is of paramount importance to enable AUV operations in these complex environments, whose continuous evolution and limited predictability pose additional challenges to marine autonomous operations. An autonomous vehicle must be able to promptly and adequately react to the challenges that a dynamic environment presents to maintain its own safety and the integrity of the precious data it collects. Despite the large interest that obstacles avoidance has received [1-3], the problem still presents open questions and areas for improvement. The project will focus on the development of robust and reliable algorithms and processes to enable an underwater vehicle to effectively detect its surrounding environment based on the information provided by onboard sensors (such as ranging or imaging sonars), and to use that knowledge to guarantee its safety to successfully complete autonomous missions. The candidate will be required to contribute to and develop innovative solutions to tackle the diverse aspects of the problem.
Providing the capabilities required to allow an underwater vehicle to safely navigate in complex and unstructured environments while avoiding obstacles can be defined as a multi-phase process requiring the vehicle to autonomously analyse the data it receives from a set of heterogeneous sensors, use that information to build awareness about its surrounding environment, and to finally react to the presence of obstacles. The candidate will be required to explore and develop innovative solutions to tackle the issues related to the diverse phases. For instance, they will develop algorithms able to learn and determine advanced sensor models to improve a vehicle’s autonomous detection capabilities, create efficient solutions to build advanced environment representations based on the limited knowledge provided by onboard sensors, and derive strategies allowing a vehicle to effectively avoid obstacles based on factors such as risk, potential reward, and a-priori knowledge of the mission type and constraints. The developed solutions will be tested in simulation first and then will be integrated on board the AUVs that the NOC is developing for under ice operations and tested in suitable field tests campaigns.
The INSPIRE DTP programme provides comprehensive personal and professional development training alongside extensive opportunities for students to expand their multi-disciplinary outlook through interactions with a wide network of academic, research and industrial/policy partners. The student will be registered at the University of Southampton and hosted at the National Oceanography Centre. Specific training will include:
- Marine robots and autonomous underwater vehicles
- Familiarisation with the assets underwater vehicles are commonly equipped with
- Underwater perception, sensors data processing, filtering, sensor fusion, optimisation, and autonomous learning
- Development of high-performance software algorithms to be deployed on real robots
- Collaborations with experts in marine technology and ocean sciences, as well as work in research groups
- Participation in field experiments and science cruises (depending on availability and schedule)
 S.M. Schillai, S.R. Turnock, E. Rogers, and A.B. Phillips, "Experimental analysis of low-altitude terrain following for hover-capable flight-style autonomous underwater vehicles", Journal of Field Robotics, vol. 36, issue 8, pp. 1399-1421, 2020.
 M.S. Wiig, K.Y. Pettersen, and T.R. Krogstad, "A 3D reactive collision avoidance algorithm for underactuated underwater vehicles”, Journal of Field Robotics, 2020.
 R.S. McEwen, S.P. Rock, and B. Hobson, “Iceberg Wall Following and Obstacle Avoidance by an AUV", Proceedings of IEEE/OES 2018 Autonomous Underwater Vehicles (AUV), Porto (PT), 2018.