Department of Engineering Science, Parks Road, Oxford, OX1 3PJ The Mobile Robotics Group (MRG) at the Oxford Robotics Institute (ORI) in the Department of Engineering Science at the University of Oxford is seeking exceptional candidates for a full-time Postdoctoral Research Assistant (PDRA) position. This EPSRC-funded position is fixed-term until October 2026. The Role You will conduct research at the intersection of perception, learning, and navigation to enable robust robot autonomy in complex, real-world environments. The post sits within our EPSRC Programme Grant in Embodied Intelligence and will advance the state of the art in localisation and scene understanding under challenging conditions. The work spans core methods, research and system-level deployment, with application domains including logistics, agriculture, social care, and inspection. About You You hold, or are close to completing, a PhD in robotics, robot learning, or a closely related field. You possess strong expertise in deep learning and robot navigation, with hands-on experience in deploying navigation algorithms and machine learning models on physical robot platforms. We are particularly interested in candidates with expertise in generative AI and curriculum learning applied to robotics, as well as in understanding and leveraging internal representations in deep-learning models, and in dataset design and curation for embodied tasks. You have a solid publication record, are conversant with the relevant literature, and are proficient in Python and common ML frameworks. What you will do You will design and evaluate learning-based methods for navigation and perception, develop and analyse representations for robust localisation and scene understanding, curate and scale datasets for embodied learning, and transition algorithms from simulation to real-world robot systems. You will collaborate closely within ORI and across the Programme Grant, contributing to the development of tools and high-quality publications, and engaging with application partners as appropriate.