Junfeng Ren

I am a master’s student at the Southern University of Science and Technology (SUSTech), working on 3D perception and scene understanding for autonomous driving and embodied agents.

My research aims to build efficient and reliable 3D perception systems that enable autonomous agents to understand, communicate, and reason about complex dynamic environments. I am particularly interested in semantic occupancy prediction, collaborative perception, token-based spatial representations, and spatio-temporal modeling.

Currently, my research focuses on efficient and predictive 3D scene understanding for autonomous agents. I study communication-efficient collaborative occupancy prediction, including adaptive token selection, spatio-temporal memory, and communication-aware fusion under limited bandwidth. I am also exploring occupancy world models that extend occupancy-based perception toward temporal reasoning and future scene prediction.

I am preparing for Ph.D. applications for Fall 2027 and seeking research assistant opportunities for Summer 2026.