Professor
Department of Mathematics
University of California, Los Angeles
Lab: Slichter Hall 3860
Office: MS 7619E
chenfanfu.jiang-at-gmail.com
UCLA Artificial intelligence & Visual Computing Laboratory
Research interest: visual computing, physical AI, 3D generative AI, embodied AI, robotics.
Publications | CV | Google Scholar | Physics-Based Simulation (Book)
My current research centers on the intersection of physical AI and visual computing.
My group builds the algorithms, simulators, and learning systems that let intelligent agents perceive, reason, and act in the physical world. We extend generative AI with spatial, temporal, and physical understanding, so models handle contact and deformation, obey physical constraints, and transfer to real embodied systems.
-
Simulation.
Robust, accurate, and scalable 3D dynamics—contact, deformation, fracture, grains—on GPUs;
differentiable pipelines and digital‑twin environments.
IPC · Embedded IPC · JGS2 · ABD · MLS‑MPM · AnisoMPM · Multi‑GPU MPM · XPBI -
Generation.
Multimodal generative models, neural rendering, and physically grounded world models that produce large‑scale
3D/4D synthetic data for training and evaluation.
Dress‑1‑to‑3 · ARM · PhysGaussian · Atlas3D · VR‑GS · PhysAnimator · ElastoGen · PhysMotion -
Robot Learning.
Training and validation stacks for contact‑ and tactile‑rich robot policies, with principled
sim‑to‑real transfer and closed‑loop benchmarks.
Taccel · GRIP · Right‑Side‑Out · Data Glove · Mirroring · BAT
Goal: close the loop between pixels, physics, and policy.