Welcome to the QUT Centre for Robotics, Marine Robotics Seminar Series — an initiative to bring together researchers, engineers, and students working at the intersection of marine science and robotics. With marine ecosystems under urgent and increasing pressure, it's never been more critical to accelerate the development and deployment of innovative technologies that can monitor, protect, and restore our oceans. Our sessions span a range of topics, from underwater computer vision and SLAM, to simulation tools, hardware design, and field deployments. Join us as we dive deep (haha) into the latest research shaping the future of marine robotics.
Watch this space for future seminars!
| Date | Speaker(s) | Affiliation | Title | Recording |
|---|---|---|---|---|
| 19/03/2026 | Dr Vincent Raoult | Griffith University | The future of marine management is now: drones, ROVs, and automated surface vessels | Not Recorded |
| 26/02/2026 | Sergei Nozdrenkov | Wildflow | Closing the Loop: From 3D Sensing to Reef Restoration | YouTube |
| Date | Speaker(s) | Affiliation | Title | Recording |
|---|---|---|---|---|
| 20/10/2025 | Dr Julie Vercelloni | Australian Institute of Marine Science | AI in coral reef monitoring: The good, the bad and the future | YouTube |
| 01/09/2025 | Dr Gideon Billings | Australian Centre for Robotics (ACFR), University of Sydney | Autonomous Underwater Systems for The Exploration of Extreme Ocean Environments | YouTube |
| 31/07/2025 | Dr Hui Sheng Lim | CSIRO | Marine autonomy: From algorithm to observations | YouTube |
| 15/05/2025 | Jonathan Sauder | EPFL | Scalable Semantic 3D Mapping of Coral Reefs with Deep Learning | YouTube |
| 13/03/2025 | Prof Rod Connolly | Griffith University | Improved monitoring of marine ecosystems with underwater computer vision | YouTube |
| 30/01/2025 | Peter Smith and Alec Tutin | QUT | Marine Robotic Simulators – Prevent Your AXV from a Watery Demise | YouTube |
Watch this space! Or send me an email at sg.raine@qut.edu.au if you don't want to miss out on future seminars.
Bio: Vincent is a marine ecologist with a broad interest in how ecological processes can be used to achieve better conservation and management of threatened communities. Vincent has strong interests in fisheries management, with a particular focus on sharks and rays. He strives to develop novel methodologies to answer or improve on numerous research issues using emerging technologies like drones and remotely operated underwater vehicles (ROVs). Vincent has studied a variety of environments, from terrestrial aquaria to coral reefs to remote seas in Tasmania.
Abstract: Our oceans are in a state of rapid and widespread change. This requires marine management to move beyond traditional, human-driven survey approaches that are costly and less efficient. In this seminar I'll discuss my experience as an ecologist using affordable modern remotely piloted tools to answer general management questions, how they compare to human-driven approaches, and their current and future integration in fisheries management and industry. I'll discuss their applications for assessing fisheries bycatch impacts, coral reef monitoring, and sea cucumber fishery management, including some integration with visual machine learning approaches.
Bio: Sergei Nozdrenkov is the founder of Wildflow, an engineering startup building the data and AI stack to model the biosphere, starting with coral reefs. Previously, he spent six years at Google, where he was a founding engineer on a biodiversity moonshot at X, Alphabet's Moonshot Factory. There, he applied machine learning to insect biodiversity, specifically focusing on pollinator health and computational pest management. His background spans large-scale distributed systems (Google Ads) and geospatial data processing (Google Maps). Sergei approaches nature through an engineering lens - aiming to increase the information bandwidth between humans and the biosphere to drive effective, scalable restoration. He is also a passionate freediver.
Abstract: While the volume of nature data is skyrocketing, we still lack the "crystal ball" needed to understand complex ecosystem dynamics and drive effective restoration. This seminar unpacks the engineering challenges of building a "digital nervous system" for the biosphere, treating reef restoration as a closed-loop control problem: Measure, Model, Act. The discussion centers on Wildflow’s technical pipeline, moving from traditional photogrammetry to high-fidelity 3D Gaussian Splatting capable of visualizing massive reef-scale datasets in the browser, and outlines the roadmap from these digital twins to multimodal foundation models. Ultimately, the goal is to bridge the gap between sensing and action, using these insights to guide targeted, localized interventions - optimizing everything from restoration site selection and coral outplanting strategies to predator control and more.
Bio: Julie’s research is at the forefront of integrating cutting-edge technologies, such as machine learning, virtual reality, and advanced Bayesian statistics to address complex challenges in coral reef conservation. She has been instrumental in projects like Virtual Reef Diver and ReefCloud, which integrate AI technology to analyse reef health for large-scale monitoring. She is at the forefront of combining AI application with advanced Bayesian statistics to analyse change and quantify the relative contribution of underlying drivers. Her work exemplifies the fusion of ecological expertise with technological innovation that breaks the barriers of ‘proof-of-concept’, providing valuable insights into the drivers of ecosystem change and informing conservation strategies.
Abstract: This talk will present recent insights from integrating AI techniques into coral reef monitoring programs, highlighting both the benefits and limitations for reef conservation, and outlining directions for future research.
Bio:
Before joining the ACFR as a Research Fellow, Gideon completed his PhD in Robotics at the University of Michigan, advised by Prof. Matthew Johnson-Roberson, followed by a short stint as a Postdoctoral Investigator at the Woods Hole Oceanographic Institution (WHOI), supervised by Dr. Richard Camilli. He is motivated in his research to develop robotic systems for the exploration of extreme and remote ocean environments, where these systems must have the capacity for high-level perception, cognition, and autonomy. He has worked with many marine robotics platforms including AUVs, gliders, and work-class ROVs, with experience in hardware and software development and field operations of these systems. His research work at the ACFR is focused on real-time perception methods to improve AUV navigation and enable autonomous underwater manipulation in both structured and unstructured environments.
Abstract:
From the deep ice shelves of Antarctica to the subsurface oceans of Europa and Enceladus, vast oceanic environments remain largely unexplored yet hold keys to understanding global climate dynamics and the potential for extraterrestrial life. While autonomous underwater vehicles can unlock access to these remote environments, developing robotic systems capable of operating in such extremes presents immense technical challenges. These platforms must rely on local sensors to interpret the scene, plan their motions, collect samples and complete other mission tasks fully autonomously, within completely unknown environments. This talk presents two research projects that advance autonomous underwater robotic capabilities for extreme environment exploration. The first project, funded by NASA, developed autonomous systems for exploring hydrothermal vent environments as terrestrial analogues for potential Europa missions. My contributions focused on perception and control algorithms that enable automated sample collection using remotely operated vehicles. This work culminated in successful field deployment at Kolumbo, an active submarine volcano in the Aegean Sea, demonstrating autonomous sample collection in the deep ocean. The second project involved modifying a Slocum Autonomous Underwater Glider to enable efficient long-range surveys beneath ice shelves. By enhancing the platform's navigation and sensing capabilities, we developed a self-sufficient system capable of extended autonomous operation in challenging under-ice environments.
Bio:
Hui Sheng is a Hobart-based research engineer with a PhD in marine engineering. He specialises in developing autonomous marine systems for ocean observations, currently working with the Autonomous Sensors Future Science Platform at CSIRO. With a strong background in engineering and computational research in the maritime sector, Hui Sheng has a track record in developing and optimising AI applications for crewed and uncrewed marine systems, as well as modelling and simulating marine vehicles and ocean environments. Hui Sheng has led and contributed to a range of field deployments, algorithm development efforts, and collaborative research projects, with a focus on real-world impact and sustainable ocean observations.
Abstract:
The demand for timely, high-resolution ocean data is becoming increasingly critical, driven by climate change, the need for improved resource management and environmental monitoring of extreme events. Autonomous marine systems are emerging as a transformative solution. These systems enabling persistent, wide-area observations with reduced human input, offering new capabilities for real-time, adaptive decision-making at sea. In this talk, I’ll share recent developments in marine robotics from CSIRO’s Autonomous Sensors Future Science Platform, focusing on mission planning algorithms and field-ready autonomy. Furthermore, it will cover the development of adaptive planning techniques, including variants of Monte Carlo Tree Search, evaluated against baseline behaviours across different environmental scenarios. This work is embedded within broader system development, including vehicle integration, onboard autonomy, and fleet operations. I will also highlight science-driven field deployments where these systems can be applied, such as autonomous surveys for water quality monitoring and fish stock assessments. Through this integrated effort, we aim to advance scalable, adaptive solutions for real-time decision-making in complex oceanographic settings.
Bio:
Jonathan Sauder is a PhD student at the École Polytechnique Fédérale de Lausanne (EPFL) in Lausanne, Switzerland working on the intersection of 3D computer vision, machine learning and coral reefs. Jonathan's research explores how recent breakthroughs in machine learning and computer vision can be used to create the next generation of coral monitoring tools. Previously, during his MSc at TU Berlin in computer science and BSc at the University of Potsdam, Jonathan worked on research in the intersection of machine learning, compressed sensing, and computer vision.
Abstract:
In light of the existential threat to coral reefs worldwide due to human activity, innovative monitoring strategies are demanded that are efficient, standardized, scalable, and economical. Mapping coral reefs from underwater videos in real time with consumer hardware could lead to the next generation of coral monitoring tools. However, coral reefs pose challenging conditions for 3D computer vision algorithms: the color degradation and blur induced by the water column, flickering caustics from sunlight hitting the wavy surface, and many moving or swaying objects (e.g. fish) often lead established systems to fail.
Self-supervised monocular SLAM emerges as a remedy, allowing to rapidly and robustly map large portions of reefs with cheap cameras without training on any ground-truth 3D information. An extension of this idea can be used to enhance existing imagery and make it suitable for conventional 3D mapping pipelines, which in turn can be used to create large-scale 3D datasets for supervised 3D foundation models. This translates to a real-world 3D semantic mapping system that is implemented in 6 countries of the Red Sea, used to analyze a steady stream of monitoring data from dozens of reef sites.
Bio:
Professor Rod Connolly is the Director of the Coast and Marine Research Centre at Griffith University and Director of the Global Wetlands Project. He is an experienced marine scientist using automated monitoring technologies to measure fish abundance and ecosystem health on reefs, seagrass meadows, saltmarshes and mangroves. Rod’s key research interests span deep learning analysis of marine imagery; ecology of coral reefs, seagrass and mangroves; and effects of climate variability and pollution on marine ecosystems.
Bio:
Peter T. Smith is a Senior Research Engineer in the REF-RAS team at QUT and a co-founder of Blind Mystics – a video game and software contracting company. Peter brings background skills in a unique blend of software engineering and video game development to marine robots. His area of expertise is robotic simulators and communication protocols, but his skills extend to all areas of marine autonomy.
Alec Tutin – The other co-founder and shadowy figure who lurks in the dark.
Abstract:
Robots are hard. Throwing robots into the sea? Now you’re just asking for trouble. You had best be sure it's designed well before you do that. Marine simulators can help you spot and fix trouble before your expensive and time-consuming field trip days; this seminar will discuss what they are good at, what they’re not, and how to get started.