Portrait of Roger Hsiao

Roger Hsiao

Graduate Researcher

UC Berkeley

I am a 5th-year PhD student in the SLICE Lab at UC Berkeley, advised by professors Sophia Shao and James Demmel. Prior to my graduate studies, I earned my B.S. degree in computer engineering from the University of Michigan–Ann Arbor, where I worked with professor David Blaauw on a long-term research project on tracking monarch butterfly migration using millimeter-scale computers. For the past two summers, I have interned at NVIDIA on the robotics perception team.

Research

Much of my prior research experience lies in designing algorithms for robotics and computer vision applications on specialized computing systems, such as GPUs, FPGAs, or custom silicon chips. My current research focuses on efficient 3D/4D reconstruction using 3D Gaussian Splatting representations and transformer-based models.

Publications

  1. 3DGS2-TR: Scalable Second-Order Trust-Region Method for 3D Gaussian Splatting

    Roger Hsiao, Yuchen Fang, Xiangru Huang, Ruilong Li, Hesam Rabeti, Zan Gojcic, Javad Lavaei, James Demmel, Sophia Shao

    ICML 2026 arXiv:2602.00395

  2. MAVERIC: A Heterogeneous Robotics SoC with 4 CPU Cores and 13 INT8/FP32 Accelerators in 16-nm-Class Technology

    Seah Kim, Jerry Zhao, Roger Hsiao, Yufeng Chi, Vighnesh Iyer, Vikram Jain, Borivoje Nikolic, Yakun Sophia Shao

    IEEE JSSC, April 2026

  3. MAVERIC: A 16nm 72 FPS, 10 mJ/frame Heterogeneous Robotics SoC with 4 Cores and 13 INT8/FP32 Accelerators

    Seah Kim, Jerry Zhao, Roger Hsiao, Yufeng Chi, Vighnesh Iyer, Vikram Jain, Borivoje Nikolic, Yakun Sophia Shao

    VLSI, June 2025 Best Student Paper Finalist

  4. SuperNoVA: Algorithm-Hardware Co-Design for Resource-Aware SLAM

    Seah Kim, Roger Hsiao, Borivoje Nikolic, James Demmel, Yakun Sophia Shao

    ASPLOS, April 2025