Mingda Xu

Mingda Xu is a Research Fellow at the Australian National University (ANU), working in Prof. Stephen Gould's group in collaboration with Seeing Machines Ltd. His research focuses on autonomous systems, combining machine learning, computer vision, and robotics. He holds a PhD in Robotics from Queensland University of Technology (2019–2022) and a Master's in Mathematics from the University of New South Wales (2016–2018).

His research interests include unsupervised learning, optimal control, differentiable optimization, and visual localization. Notable projects involve feature-shaping methods for anomaly detection, diffusion models for pose estimation, and model-based control techniques like differentiable dynamic programming. He has also explored applications in visual place recognition and SLAM (Simultaneous Localization and Mapping).

• Key projects: Unsupervised learning for long-form videos, differentiable DTW for visual place recognition, and discrete-time optimal control.

His work has been recognized with a CVPR 2024 Best Paper Award nomination. He collaborates actively with industry and academia, contributing to the development of scalable and robust autonomous systems.

He is registered to supervise research students and has contributed to multiple grants focused on advancing robotics and AI technologies.

Mingda's lab work integrates robotics, computer vision, and machine learning, emphasizing end-to-end solutions for real-world applications in autonomous systems.