01
3D/4D Robot Vision
Learning geometry, motion, semantics, and localization from cameras, LiDAR, point clouds, neural fields, and 4D observations.
Read morePIRLab.github.io
Physical Intelligence and Robotics Lab
We build robots that perceive dynamic 3D worlds, predict physical futures, and act through models grounded in geometry, semantics, and mechanics.
Mission
Robotic intelligence should understand the physical world before it moves in it.
PIRLab is led by Guangming Wang and connects research activity across Cambridge, Ireland, and Shanghai Jiao Tong University. Our work sits at the intersection of robot vision, embodied AI, world models, 3D reconstruction, and robot learning.
We focus on compact ideas with practical force: rich 3D/4D perception, predictive models of scene dynamics, and physically meaningful action policies for mobile robots, manipulation, construction automation, and digital twins.
Research
Three directions, one physical intelligence stack.
02
World-Model-Based Prediction
Building predictive models that forecast scene flow, map evolution, future observations, and action-conditioned physical change.
Read more
03
Physics-Based Robot Action
Translating perception and prediction into contact-rich manipulation, real-to-sim-to-real learning, robust estimation, and efficient control.
Read moreLead
Guangming Wang
Guangming Wang is an Assistant Professor at the School of Engineering, University of Galway, Cambridge-Ireland International Centre, and a Research Associate in the Department of Engineering at the University of Cambridge. He received his PhD from the Intelligent Robotics and Machine Vision Lab at Shanghai Jiao Tong University.
He serves as Associate Editor for IEEE RA-L, ICRA, and IROS, and was awarded the Best Associate Editor Award of ICRA 2026. His research connects robot perception, localization, mapping, world modeling, and robot manipulation.
Papers
The publication page is now a clean CV-based paper list.
Papers are listed as citations only, while media-rich selected work is shown on the Research page.
View papersIndustry Application
Three application modules for deployable physical intelligence.
Industry applications are organized into autonomous mobile robots/autonomous driving, robot brain and embodied intelligence/VLA, and industrial autonomous robots for construction automation.
View applicationsGroups
Three connected research homes.
Cambridge
Cambridge Robotics Group
3D robot perception, localization, mapping, construction automation, and embodied physical intelligence.
IrelandCambridge-Ireland International Centre Research Group
World models, predictive robot vision, physical AI, and international collaboration across Cambridge and Ireland.
SJTUShanghai Jiao Tong University Research Group
Long-running research roots in robot vision, LiDAR odometry, 3D scene flow, and manipulation learning.
Join
We welcome students who want to build physical intelligence.
PIRLab is open to highly motivated PhD, MPhil, research assistant, and visiting student applicants.
Join PIRLabNews
Recent highlights
- One paper on robot localisation accepted by ECCV 2026.
- One paper on world model accepted by ICML 2026.
- Invited talk on World Models at the ASPIRE Computer Vision Workshop at Oxford VGG.
- One SLAM paper accepted by IEEE TPAMI.
- One robot manipulation paper accepted by ICRA 2026.
- One paper on SLAM accepted by AAAI 2026.
- Invited talk on Real2Sim2Real at the Cambridge Computer Vision Workshop.
- NeRF in Robotics survey accepted by IJRR.