Athena Seminar Series: Toward Scalable, Real-Time, and AI-Native Immersive Systems

Presenter: Bo Han

: Immersive computing is emerging as a new computing paradigm that enables applications in domains such as healthcare, education, and training. However, developing practical immersive systems remains challenging due to the massive data volume of 3D content and the stringent latency and reliability requirements of interactive applications. In this talk, I will present my research on building scalable and intelligent systems for immersive computing. First, I will discuss how to make immersive content deliverable under bandwidth constraints using systems such as ViVo and NeVo, which exploit the characteristics of point clouds and neural representations to improve streaming efficiency. Next, I will present M5 and MuV$^2$, which address the challenges of supporting scalable, immersive experiences for multiple users over dynamic wireless networks. Finally, I will introduce MagicStream and LITE, which leverage semantic communication and multimodal loss recovery to enable low-latency, high-fidelity, and resilient immersive telepresence. I will conclude with a vision for reality-centric immersive computing, highlighting opportunities in human-AI interaction and humanoid teleoperation.