Jianhua Sun
Office: School of Artificial Intelligence, Room 209
Email: gothic [at] sjtu (dot) edu (dot) cn
Google Scholar / SJTU Profiles
I am an Assistant Professor in the School of Artificial Intelligence at Shanghai Jiao Tong University. I finished my Ph.D. in the Department of Computer Science at Shanghai Jiao Tong University in 2025, under the supervision of Prof. Cewu Lu. Prior to that, I received my Bachelor's degree (Computer Science) from Shanghai Jiao Tong University in 2020.
Research
My research interests lie in Physical AI --- teaching AI to learn physical models of real-world concepts in order to build machines that possess a rich understanding of the world for perceiving, reasoning, interacting with the environment, and communicating with humans. My research spans from physical world representations to learning algorithms, including
- representations of physical world concepts in an analytic form, enabling rigorous numerical computation and simulation
- structured priors that encode phsyical laws and social rules, guiding AI in commonsense reasoning
- physical world perception, generation and planning algorithms adhere to the fundamental rules of how the world operates
- lifelong learning to optimize physical-world models and evolve learning algorithms
In addition to developing the representations and algorithms themselves, we simultaneously explore their applications in various domains:
- robotics and embodied AI
- simulation and world model
- multi-modal fundation model
- autonomous driving
- AI-assisted design and content generation
Recruitment: Looking for self-motivated students (Master & Ph.D. spring & fall 2026 & later, undergraduate interns, visitors) to join us in MVIG. Feel free to email me directly with your CV (Subject: Recruitment + Institution + Field of Study + Academic Year + Name).
Publications
(* and † indicate equal contribution and corresponding author)
|
Arti-PG: A Toolbox for Procedurally Synthesizing Large-Scale and Diverse Articulated Objects with Rich Annotations |
|
Interactive Adjustment for Human Trajectory Prediction with Individual Feedback |
|
Discovering Conceptual Knowledge with Analytic Ontology Templates for Articulated Objects |
|
ConceptFactory: Facilitate 3D Object Knowledge Annotation with Object Conceptualization |
|
Modality Exploration, Retrieval and Adaptation for Trajectory Prediction Three Steps to Multimodal Trajectory Prediction: Modality Clustering, Classification and Synthesis |
|
InstaBoost++: Visual Coherence Principles for Unified 2D/3D Instance Level Data Augmentation Instaboost: Boosting Instance Segmentation via Probability Map Guided Copy-Pasting |
|
Stimulus Verification is a Universal and Effective Sampler in Multi-Modal Human Trajectory Prediction |
|
Human Trajectory Prediction with Momentary Observation |
|
Correlation Field for Boosting 3D Object Detection in Structured Scenes |
|
Unified and Fast Human Trajectory Prediction Via Conditionally Parameterized Normalizing Flow |
|
Recursive Social Behavior Graph for Trajectory Prediction |
|
Symbol-LLM: Leverage Language Models for Symbolic System in Visual Human Activity Reasoning |