Mehdi’s passion for robotics and computer vision began in 2009–2010, when he participated in International RoboCup competitions in Austria and Singapore. Motivated by these early successes, he pursued a master’s degree in Control Theory, establishing a strong foundation in the fundamentals of robotics and control.

Continuing his studies, Mehdi completed his PhD at the University of Adelaide, under the guidance of Prof. Ian Reid and in affiliation with the Australian Centre for Robotic Vision. His research merged robotics and computer vision to develop real-time, structure-aware, and object-centric semantic visual SLAM. By integrating advanced object detection techniques and leveraging robust 3D scene priors, he addressed key limitations of traditional geometric SLAM frameworks.

Today, Mehdi is a Postdoctoral Research Fellow at the Australian Institute for Machine Learning, where he contributes to projects that span Computer Vision, Robotics, and Machine Learning. His current research focuses on two main streams:

Multi-Modal Sensor Fusion and Scene Understanding (Industry Partner Collaboration)

• Stereo visual odometry
• Bird’s-eye-view (BEV) mapping
• Enhanced scene understanding for autonomous driving applications

3D Reconstruction and Neural Representations

• Exploring neural reconstruction methods, including Gaussian splatting
• Investigating advanced geometric priors for 3D neural reconstruction