Shanghai Jiao Tong University

SJTU Research Group

The Shanghai research root of PIRLab: robot perception, LiDAR odometry, 3D scene flow, robust estimation, and robot manipulation learning.

Back to PIRLab groups

Purpose

From rigorous 3D perception to robot action.

The SJTU group reflects PIRLab's long-running research foundation in the Intelligent Robotics and Machine Vision community at Shanghai Jiao Tong University.

The group emphasizes strong student mentoring, open-source robot vision systems, and publication-quality research across computer vision, robotics, AI, and intelligent transportation.

LiDAR Odometry

Efficient 3D deep odometry, point-cloud representation learning, and robust localization pipelines.

Scene Flow

Large-scale 3D motion estimation, auto-labelling, uncertainty-aware learning, and dynamic perception.

Manipulation

Reinforcement learning, base controllers, real-to-sim-to-real transfer, and physically grounded policies.

Group Members

Current SJTU members.

Guangming Wang

Group Lead

Guangming Wang

Assistant Professor at the Cambridge-Ireland International Centre and Research Associate at the University of Cambridge; robot vision, LiDAR odometry, 3D scene flow, robust estimation, and manipulation learning.

Siting Zhu

PhD Student

Siting Zhu

Research interest: semantic SLAM. Co-supervised with Prof. Hesheng Wang.

Wenhua Wu

PhD Student

Wenhua Wu

Research interest: 3D reconstruction. Co-supervised with Prof. Zhe Liu.

Yu Zheng

PhD Student

Yu Zheng

Research interest: mobile manipulation. Co-supervised with Prof. Hesheng Wang.

Yuxuan Wu

PhD Student

Yuxuan Wu

Research interest: computer vision and deep learning for robot manipulation. Co-supervised with Prof. Hesheng Wang.

Advisory Board

SJTU advisory board.

Tang Jun Yuan Chair Professor at Shanghai Jiao Tong University; intelligent robots, embodied AI, autonomous vehicles, and machine vision.

Zhe Liu

Advisor

Prof. Zhe Liu

Tenure-track associate professor at SJTU; large-scale robotic systems, resilient navigation, and robot control.

Former Members

Selected past mentees and destinations.

Chensheng Peng

PhD student at UC Berkeley.

Jiuming Liu

PhD student at the University of Cambridge.

Shijie Zhao

Master student at Georgia Tech.

Yanfeng Guo

Postgraduate at UCLA.

Xiaoyu Tian

Postgraduate at Columbia University.

Hanwen Liu

Postgraduate at TUM.

Minjian Xin

Postgraduate at UCSD.

Yuyou Zhang

PhD student at CMU.

Yunzhe Hu

PhD student at HKU.

Representative work

Core papers from the SJTU research line.

EfficientLO-Net architecture

LiDAR odometry

Efficient 3D Deep LiDAR Odometry

A projection-aware architecture for efficient point-cloud odometry, published in IEEE TPAMI.

3D scene flow network figure

3D scene flow

What Matters for 3D Scene Flow Network

Design study and architecture for high-performance 3D scene flow estimation, published at ECCV.

Sparse-reward robot manipulation learning

Robot manipulation

Learning of Long-Horizon Sparse-Reward Robotic Manipulator Tasks With Base Controllers

Safe and sample-efficient reinforcement learning for long-horizon manipulation tasks, published in IEEE TNNLS.

Mentoring

A training ground for ambitious robotics students.

Formerly mentored students have continued to leading universities and research labs including UC Berkeley, Princeton, HKU, Columbia, Georgia Tech, UCSD, UCLA, TUM, SJTU, and related robotics groups.

Contact