PhD Candidate in multi-sensor fusion and representation learning for underwater robotics

This is NTNU<p>NTNU is a broad-based university with a&nbsp;technical-scientific profile and a focus&nbsp;in professional education. The university is located in three cities with headquarters in Trondheim.</p><p>At NTNU, 9,000 employees and 43,000 students work to create knowledge for a better world.</p><p>You will find more information about working at NTNU and the application process <a href="https://www.ntnu.edu/vacancies/" rel="nofollow">here.</a></p><p>&nbsp;&nbsp;&nbsp;</p>Video: <a href="https://youtu.be/Xt-yHCN5QS0" rel="nofollow">https://youtu.be/Xt-yHCN5QS0</a>About the position<p>This PhD project focuses on developing novel methods for multi-sensor fusion and representation learning that exploit remote sensing priors - satellite optical and multispectral imagery, synthetic aperture radar (SAR), aerial photogrammetry, and prior bathymetry - to augment what an underwater robot can perceive,&nbsp;localize&nbsp;within, and reason about. The aim is to advance principled cross-modal and cross-scale representation learning for the underwater domain, and to&nbsp;demonstrate&nbsp;that richer, uncertainty-aware environmental models lead to more capable and efficient autonomous underwater&nbsp;surveys.&nbsp;</p><p>This PhD project will develop principled methods for fusing these heterogeneous observation streams into coherent, uncertainty-aware environmental representations that an underwater robot can exploit for perception, habitat mapping, and adaptive survey planning. The central scientific challenge is cross-modal and cross-scale representation learning: how to encode geometric, acoustic, optical and environmental observations made at different scales, resolutions, modalities and times into a shared representation that is actionable for an onboard robot.&nbsp;Applications include benthic habitat mapping (cold-water coral reefs, kelp forests, seafloor geology), infrastructure inspection and persistent environmental monitoring. The project will investigate whether and how recent foundation models for earth observation and underwater perception can be adapted and extended to support this multi-modal, cross-scale fusion challenge.</p><p>Bridging the information gap between broad-scale remote sensing and close-range underwater robot&nbsp;perception&nbsp;represents&nbsp;one of the most compelling open problems in marine robotics. This PhD project focuses on developing novel methods for multi-sensor fusion and representation learning that exploit remote sensing priors, including satellite optical and multispectral imagery, synthetic aperture radar (SAR), aerial photogrammetry, acoustic bathymetry, synthetic aperture sonar and side-scan sonar, to augment the local sensors that an underwater robot uses to perceive,&nbsp;localize&nbsp;within, and reason about. The aim is to advance principled cross-modal and cross-scale representation learning for the underwater domain, and to&nbsp;demonstrate&nbsp;that richer, uncertainty-aware environmental models lead to more capable and efficient autonomous underwater&nbsp;surveys.&nbsp;</p><p>The PhD candidate will be supervised by Professor Oscar Pizarro at the&nbsp;<a href="https://www.ntnu.edu/imt/marine-structures" rel="nofollow">Department of Marine Technology (IMT)</a>, and the position is part of the&nbsp;recently funded&nbsp;Norwegian Centre for Embodied AI (NCEI).&nbsp;</p><br>About the project<p><strong>Join a nation-wide team:&nbsp;</strong><a href="https://www.ntnu.edu/ncei" rel="nofollow">The Norwegian Centre for Embodied AI (NCEI),</a> one of Norway's six national AI centers, is recruiting outstanding researchers to advance a universal science of embodied intelligence. NCEI brings together leading robotics and AI groups with key partners from industry and the public sector to study how intelligence&nbsp;emerges&nbsp;from the interaction between body, computation, and environment, across flying, ground, and aquatic robot configurations. Our mission is to chart a generalizable path for physical AI and transform how robot morphology and autonomy are co-designed, enabling new generations of systems tailored to their operational environments and missions. Successful candidates will join an international community with world-class facilities and strong collaborations across Norwegian universities, research institutes, industry, public agencies, and leading global institutions. We welcome motivated applicants in robotics, control, AI, machine learning, physics, and related fields, including early-stage researchers eager to contribute to this emerging scientific frontier.&nbsp;<strong>&nbsp;</strong></p><br>Duties of the position<ul><li>Complete your doctoral education leading to the PhD degree.&nbsp;</li><li>Conduct and publish research of high quality within the framework described above.&nbsp;</li><li>Develop fundamental contributions in multi-sensor fusion and representation learning for underwater robotics, with an emphasis on exploiting remote sensing priors.&nbsp;</li><li>Develop implementable methods for onboard robot&nbsp;perception,&nbsp;localization, and adaptive survey planning.&nbsp;</li><li>Conduct experimental deployments and field evaluation using NTNU's Fjordlab AUVs and other infrastructure with acoustic (MBES, SAS), visual, and hyperspectral sensing payloads.&nbsp;</li><li>Participate in international activities such as conferences and/or research stays abroad.&nbsp;</li><li>Collaborate with other researchers within the department, and across departments at NTNU.&nbsp;</li><li>Supervise master's&nbsp;thesis&nbsp;students related to the project.&nbsp;</li></ul><p>Be prepared for changes to your work duties after employment.</p><br>Required selection criteria<ul><li>You must have a relevant master's degree in marine technology, cybernetics, mechatronics, control systems, computer science or equivalent, with strong training in robotics, computer vision, state estimation, machine learning or statistical signal processing</li><li>Your course of study must correspond to a five-year Norwegian course, where 120 credits have been obtained at master's level. Master students graduating summer 2026 are eligible to apply.</li><li>You must have a strong academic background from your previous studies and have an average grade from your Master's degree study, or equivalent education, which is equal to B or better compared to&nbsp;<a href="https://i.ntnu.no/wiki/-/wiki/English/Grading+scale" rel="nofollow">NTNU's grading scale</a>. If you do not have letter grades from previous studies, you must have an equally good academic foundation. If you have a weaker grade background, you may be considered if you can document that you are particularly suitable for a PhD education.</li><li>You must meet the requirements for admission to the&nbsp;<a href="https://www.ntnu.edu/studies/phiv%29" rel="nofollow">faculty’s Doctoral Programme in Engineering</a></li><li>Good oral and written presentation skills in English</li></ul><p>PLEASE NOTE: For detailed information about what the application must contain, see paragraph “About the application”.</p><p>The appointment is to be made in accordance with <a href="https://i.ntnu.no/wiki/-/wiki/English/NTNUs+guidelines+for+recruitment+positions+" rel="nofollow">NTNUs guidelines for recruitment positions</a> for general criteria for the position.</p><br>Preferred selection criteria<ul><li>Solid theoretical background in robot&nbsp;perception, navigation and mapping.</li><li>Background in one or more of: sensor fusion, probabilistic state estimation, underwater&nbsp;perception, remote sensing image analysis, or 3D reconstruction&nbsp;</li><li>Deep understanding of modern machine learning including foundation models and/or self-supervised&nbsp;representation&nbsp;learning&nbsp;</li><li>Solid programming skills in Python and C++. Experience with ROS2 is a plus</li><li>Experience with simulators for robotics, geospatial data tools and remote sensing processing</li><li>Strong skills i