Zicheng Zeng (曾子乘)
Zicheng Zeng (曾子乘)

PHD Student of Computer Science

I am a first year PHD student at Shanghai Qi Zhi Institute and VDI, ShanghaiTech University, advised by Prof. Li Yi and Prof. Jiayuan Gu. Previously, I obtained my bachelor’s degress of engineering majored in Artificial Intelligence from South China University of Technology, advised by Prof. Huaidong Zhang.

I am currently interning at Galbot. My research goal is to build human-like visual intelligence systems that could interact with environments as adaptive cognitive agents. My current research focus on perceptive humanoid locomotion and navigation.

Download CV
Interests
  • Embodied AI
  • Humanoid robot learning
  • 3D Computer vision
Education
  • B.Eng. in Artificial Intelligence

    South China University of Technology, 2026

News
  • 2026-09 One paper accepted to IROS 2026.
  • 2026-06 One paper accepted to IEEE RA-L (transfer to ICRA 2027).
  • 2026-02 One paper accepted to CVPR 2026 Findings.
  • 2024-12 I attended the CSIG 2024 conference, deeply grateful for the support of Prof. Huaidong Zhang.
  • 2024-07 I was honored to be invited to participate in the YPEC 2024 at City University of Hong Kong, where I had the opportunity to present the poster showcasing our project.
Publications

\* indicates equal contribution, † indicates the corresponding author.

Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data
Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data
IROS 2026

Zhikai Zhang * , Haofei Lu * , Yunrui Lian * , Ziqing Chen , Yun Liu , Chenghuai Lin , Han Xue , Zicheng Zeng , Zekun Qi , Shaolin Zheng , Qing Luan , Jingbo Wang , Junliang Xing , He Wang , Li Yi

We present LATENT, a framework that learns athletic humanoid tennis skills from imperfect human motion data, enabling the Unitree G1 robot to sustain multi-shot rallies with human players.

Collision-Free Humanoid Traversal in Cluttered Indoor Scenes
Collision-Free Humanoid Traversal in Cluttered Indoor Scenes
RA-L 2026

Han Xue * , Sikai Liang * , Zhikai Zhang * , Zicheng Zeng , Yun Liu , Yunrui Lian , Jilong Wang , Qingtao Liu , Xuesong Shi , Li Yi

We propose Humanoid Potential Field (HumanoidPF), a representation that tightly bridges environmental perception with whole-body control, enabling humanoids to hurdle, crouch, and squeeze through cluttered indoor scenes, along with a Click-and-Traverse teleoperation system for single-command traversal.

Switch-JustDance: Benchmarking Whole-Body Motion Tracking Policies Using a Commercial Console Game
Switch-JustDance: Benchmarking Whole-Body Motion Tracking Policies Using a Commercial Console Game
CVPR 2026 Findings

Jeonghwan Kim * , Wontaek Kim * , Yidan Lu * , Jin Cheng * , Fatemeh Zargarbashi * , Zicheng Zeng * , Zekun Qi * , Zhiyang Dou , Nitish Sontakke , Donghoon Baek , Sehoon Ha , Tianyu Li

We present Switch-JustDance, a low-cost and reproducible benchmarking pipeline that leverages the motion-sensing console game Just Dance on the Nintendo Switch to evaluate whole-body humanoid motion tracking controllers through the game's built-in scoring system.