FusionSense: Bridging Common Sense, Vision, and Touch for Robust Sparse-View Reconstruction
Vision, touch, and common-sense priors for reconstructing objects from sparse observations.
Robotics · NYU CREO
Jing Zhang 张婧
Self-Improving Robots
Perception, assembly, and learning through physical interaction.
I am a Faculty Fellow at NYU’s Center for Robotics and Embodied Intelligence (CREO). My research brings together robot learning, multimodal perception, and assembly reasoning.
My long-term goal is to develop robots that can assemble, test, and improve robotic systems. My current work develops foundations for this vision through vision and touch, reasoning about incomplete 3D structures, and learning to interact with the physical world.
I am on the faculty job market in robotics and embodied AI.
Toward Self-Improving Robots
Research visionHow can robots build better robots? I aim to connect understanding parts, physical assembly, and functional testing, so that the outcomes of one iteration guide the next.
- 01 · Current foundations
Understand parts
Combine multimodal perception with reasoning about incomplete shapes and their relationships.
Perception / Assembly reasoning - 02 · Future direction
Assemble robots
Turn structural reasoning into physical skills for manipulating parts, assembling systems, and recovering from errors.
- 03 · Long-term goal
Test and improve
Use functional tests to improve designs, assembly strategies, and control policies across successive iterations.
The loop: understand → assemble → test → improve → build again.
Selected Publications
Full list on Google Scholar ↗* Equal contribution. † Corresponding author.
CRAG: Can 3D Generative Models Help 3D Assembly?
Jointly assembling observed parts and generating missing geometry to reason about incomplete objects.
GARF: Learning Generalizable 3D Reassembly for Real-World Fractures
Learning generalizable 3D reassembly for real-world fractures from synthetic training data.
Wanderland: Geometrically Grounded Simulation for Open-World Embodied AI
Geometrically grounded simulation built from real environments for open-world embodied AI.
CityWalker: Learning Embodied Urban Navigation from Web-Scale Videos
Learning urban navigation from thousands of hours of human walking and driving videos.
LUWA Dataset: Learning Lithic Use-Wear Analysis on Microscopic Images
An open dataset for studying how foundation models interpret microscopic traces on ancient tools.
Selected Coverage
NYU Tandon · September 15, 2026
Teaching AI to Make Sense of the Physical World
A faculty profile on my research connecting robotics, simulation, and reasoning about incomplete physical environments.
Read the profile ↗NYU Tandon · October 13, 2026
Building AI to Piece Together Broken Archaeological Finds
Our Schmidt Sciences-supported collaboration develops AI tools for reconstructing fragmented artifacts and bones.
Read the story ↗Research Leadership & Funding
DigitizedNYC
Large-Scale 4D Urban Digital Twin for Embodied AI. Real-world environments for robot simulation, learning, and evaluation.
NSF OAC · Award #2514030 · $599,806 total project award
Fragmented Legacy
Turning broken 3D material evidence into testable archaeological hypotheses. A collaboration on reconstructing incomplete objects with AI.
Humanities and AI Virtual Institute · $805,556 total project budget over 3 years · Funder announcement ↗
Functional Dexterous In-Hand Reorientation
Enabling functional dexterous in-hand reorientation for general-purpose robotic manipulation.
NYU Postdoctoral Research and Professional Development Grant · 2025
Teaching & Mentoring
Courses at NYU
- Fall 2026Mathematics for Robotics
ROB-GY 6013 · Instructor
- Fall 2025Robot Perception
Instructor
- Spring 2025Robot Vision
Co-instructor
Student research
I advise students through NYU ARISE and the NYU Self Drive VIP team.
Research mentorship across robotics, perception, and interdisciplinary AI.
Recent News
- NYU featured our Fragmented Legacy project, supported by Schmidt Sciences.
- I joined NYU CREO as a Faculty Fellow and am teaching Mathematics for Robotics.
- CRAG was accepted to ICML 2026.
- Wanderland (Highlight) and Thinking in 360° were accepted to CVPR 2026.
- I was selected for MIT EECS Rising Stars 2025.
Background
Previously, I was a postdoctoral researcher with AI4CE, led by Chen Feng, and the Anthrotopography Lab, led by Radu Iovita, at NYU. I received my Ph.D. in Photogrammetry and Remote Sensing from Wuhan University in 2023, advised by Bin Luo and Yajun Wang, after completing my B.E. (2017) and M.S. (2019) there.
Academic service & selected talks
I review for ICRA, CVPR, IEEE Robotics and Automation Letters, Scientific Reports, PLOS, Optics and Lasers in Engineering, Optics Letters, and Optics Express.
- Invited talk, CVPR Workshop on Spatial Intelligence for Cultural Heritage, June 2025.
- Invited talk, Paleoanthropology Society Annual Meeting, April 2025.
- Invited talk, NYU Abu Dhabi Institute, January 2025.
Outside research, I enjoy dancing.