School of Computing, Engineering and Technology

Vision-Based Autonomous Navigation for Sustainable Robotics in Marine Environment

girl on a mountain looking at the sun rise

This research aims to enhance autonomous underwater navigation by integrating vision-based sensor fusion and AI control. Focusing on complex marine environments, it will develop high-resolution environmental models using camera and sonar data, ultimately promoting efficient and sustainable exploration of underwater ecosystems through improved navigation and data collection methods.

Award

MRes | MPhil | PhD | MSc by Research

Start Date

Usually February and October - at individual School's discretion

Research Opportunities

Summary

This proposed research seeks to enhance autonomous navigation for robotics in marine environments by integrating vision-based sensor fusion with AI-driven robotic control systems. The primary objective is to develop effective methodologies that enable autonomous underwater robots (AURs) to navigate successfully, even in challenging conditions such as limited visibility, complex seafloor topographies, and varying environmental factors. Traditional sonar-based navigation techniques often lack the precision for detailed, real-time underwater mapping. As a result, this project’s emphasis on advanced vision-based solutions is both timely and impactful.

The project seeks to integrate visual data from cameras and sonar sensors to generate precise, real-time 3D models of underwater environments. This data will be processed using AI-driven algorithms designed to interpret depth, identify objects, and detect environmental changes. These algorithms will be closely linked with robotic control systems to facilitate seamless and adaptive navigation capabilities. The research will focus on establishing reliable sensor fusion strategies that effectively work in tandem with control mechanisms, ensuring real-time navigation and obstacle avoidance.

This research will be executed in three main phases:

The first phase will focus on capturing synchronised visual and sonar data. High-precision calibration will align data from camera and sonar sensors, creating a solid foundation for subsequent algorithmic processing and integration.
In the second phase, machine learning models will be developed to enhance the robot’s ability to interpret its environment. This will facilitate depth estimation and object identification, enabling the construction of accurate and dynamic 3D maps—crucial for navigating and adapting to underwater conditions.
Finally, the last phase will concentrate on developing integration strategies that connect AI-driven environmental understanding with robotic control systems. This will ensure that the robot can navigate autonomously and respond in real-time to underwater obstacles and terrain.

Through its innovative approach to sensor fusion and AI in underwater environments, this research aims to advance sustainable marine exploration and environmental monitoring. Strengthened robotic capabilities in data collection and navigation will support sustainable practices in the marine industry and provide valuable insights for ocean conservation efforts.

The ideal candidate for this research should have a background in similar fields such as robotics, computer vision, or artificial intelligence and possess programming skills (C++, Python & MATLAB, OpenCV, Open3D) and experience with machine learning frameworks (e.g., TensorFlow or PyTorch). Familiarity with oceanography or underwater technology would be beneficial but is not required. The project will provide comprehensive training in sensor fusion, data analysis, and control integration.