Portrait of Keshu Wu

Keshu Wu

Postdoctoral Research Associate
Texas A&M University
About Me

Greetings! I am Keshu Wu, a Postdoctoral Researcher at Texas A&M University, working at the intersection of intelligent transportation, artificial intelligence, and cyber-physical systems.

I received a Ph.D. degree in Civil and Environmental Engineering from the University of Wisconsin–Madison. I also hold an M.S. degree in Computer Sciences from the University of Wisconsin–Madison, and an M.S. degree in Civil and Environmental Engineering from Carnegie Mellon University. I received a B.E. degree in Civil Engineering from Southeast University.

Research Overview

My research focuses on developing interactive and trustworthy intelligent systems that operate within complex physical environments. These systems aim to enable safe and coordinated interactions among humans, infrastructure, and autonomous agents, supporting the next generation of intelligent mobility and infrastructure systems.

My research specifically involves:

1) Connected and Automated Mobility Intelligence: modeling interactions among vehicles, infrastructure, and humans, and developing AI methods for safe motion prediction, planning, and control within connected vehicle–road–cloud and V2X systems.
2) Digital Twin and Cyber-Physical Systems: building end-to-end digital twin pipelines that integrate data, simulation, and system optimization, while leveraging generative AI and LLM-based reasoning to synthesize scenarios and agent behaviors.
3) Spatiotemporal AI for Networked Systems: developing methods for spatiotemporal reconstruction, forecasting, and scalable spatial–temporal network analysis to enable data fusion and system understanding across large-scale dynamic networks.

My research integrates methods from machine learning, generative AI, network science, and computational optimization with domain knowledge in transportation engineering and cyber-physical systems. Through this interdisciplinary approach, I aim to develop intelligent infrastructure and mobility systems that improve transportation safety, efficiency, and everyday human mobility.

Overview of Keshu Wu's research: connected and automated mobility intelligence, digital twins and cyber-physical systems, and spatiotemporal AI for networked systems.
Experience
  • Aug. 2024 - Present
    Texas A&M University
    Texas A&M University
    Postdoctoral Research Associate
  • Sep. 2019 - Jun. 2024
    University of Wisconsin-Madison
    University of Wisconsin-Madison
    Research Assistant
Education
  • Sep. 2019 - Jun. 2024
    University of Wisconsin-Madison
    University of Wisconsin-Madison
    Ph.D. in Civil and Environmental Engineering
    M.S. in Computer Sciences
  • Sep. 2017 - Dec. 2018
    Carnegie Mellon University
    Carnegie Mellon University
    M.S. in Civil and Environmental Engineering
  • Sep. 2013 - Jul. 2017
    Southeast University
    Southeast University
    B.E. in Civil Engineering
News
  1. 2026
  2. Sep I will give a lectern presentation at MIDTRANS 2026 in Lawrence, KS.
  3. Jan I will present 8 research posters at TRBAM 2026 in Washington, D.C.
  4. 2025
  5. Jan I will present 1 podium talk and 5 research posters at TRBAM 2025 in Washington, D.C.
  6. 2024
  7. Aug I will be joining CEE and LAUP at Texas A&M University as a postdoc.
  8. Jun I will present 1 podium talk at ASCE ICTD 2024 in Atlanta, GA.
  9. May I received my Ph.D. degree from the UW-Madison.
  10. 2023
  11. Jun I will present 1 podium talk at ASCE ICTD 2023 in Austin, TX.
Selected Publications (view all )
AURORA: A Natural Language–Driven Agentic Framework for Understanding, Reasoning, and Orchestrating Reliable Air–Ground Co-SimulationUnder Review
AURORA: A Natural Language–Driven Agentic Framework for Understanding, Reasoning, and Orchestrating Reliable Air–Ground Co-Simulation

Wu, K., Zhang, H., Gan, R., Gao, X., Li, X., Tu, Z., Zhou, Y.

Under Review (2026)

[Website]

# LLM# air-ground collaboration# agentic framework# co-simulation# UAV

Under Review
AURORA: A Natural Language–Driven Agentic Framework for Understanding, Reasoning, and Orchestrating Reliable Air–Ground Co-Simulation

Wu, K., Zhang, H., Gan, R., Gao, X., Li, X., Tu, Z., Zhou, Y.

Under Review 2026

[Website]

# LLM# air-ground collaboration# agentic framework# co-simulation# UAV

Learning Higher-Order Structure from Incomplete Spatiotemporal Data: Multi-Scale Hypergraph Laplacians with Neural RefinementarXiv
Learning Higher-Order Structure from Incomplete Spatiotemporal Data: Multi-Scale Hypergraph Laplacians with Neural Refinement

Wu, K., Li, S., Li, Z., Fan, Z., Li, X., Zhou, Y.

Under Review (2026)

[Preprint]

# spatiotemporal imputation# hypergraph learning# higher-order structure

arXiv
Learning Higher-Order Structure from Incomplete Spatiotemporal Data: Multi-Scale Hypergraph Laplacians with Neural Refinement

Wu, K., Li, S., Li, Z., Fan, Z., Li, X., Zhou, Y.

Under Review 2026

[Preprint]

# spatiotemporal imputation# hypergraph learning# higher-order structure

Knowledge Is Not Static: Order-Aware Hypergraph RAG for Language ModelsarXiv
Knowledge Is Not Static: Order-Aware Hypergraph RAG for Language Models

Wu, K., Kuai, C., Li, Z., Jiang, J., Shen, S., Wang, S., Hu, C., Tu, Z., Zhou, Y.

Under Review (2026)

[Preprint]

# knowledge graph# hypergraph# RAG# LLM

arXiv
Knowledge Is Not Static: Order-Aware Hypergraph RAG for Language Models

Wu, K., Kuai, C., Li, Z., Jiang, J., Shen, S., Wang, S., Hu, C., Tu, Z., Zhou, Y.

Under Review 2026

[Preprint]

# knowledge graph# hypergraph# RAG# LLM

A Deep Learning Enabled Economical Informatization Framework with Sparsely Located SensorsUnder Review
A Deep Learning Enabled Economical Informatization Framework with Sparsely Located Sensors

Wu, K., Zhou, Y., Li, X., Ye, X., Tu, Z.

Under Review (2026)

# inductive learning# hybrid sensor network# graph transformer

Under Review
A Deep Learning Enabled Economical Informatization Framework with Sparsely Located Sensors

Wu, K., Zhou, Y., Li, X., Ye, X., Tu, Z.

Under Review 2026

# inductive learning# hybrid sensor network# graph transformer

V2XSynth: An LLM-Driven, Retrieval-Augmented Framework for Realistic V2X Scenario SynthesisUnder Review
V2XSynth: An LLM-Driven, Retrieval-Augmented Framework for Realistic V2X Scenario Synthesis

Wu, K., Zhang, H., Li, P., Gan, R., You, J., Tu, Z., Zhou, Y.

Under Review (2026)

[Website]

# LLM# RAG# V2X# scenario generation

Under Review
V2XSynth: An LLM-Driven, Retrieval-Augmented Framework for Realistic V2X Scenario Synthesis

Wu, K., Zhang, H., Li, P., Gan, R., You, J., Tu, Z., Zhou, Y.

Under Review 2026

[Website]

# LLM# RAG# V2X# scenario generation

AI²-Active Safety: AI-enabled Interaction-aware Active Safety Analysis with Vehicle DynamicsarXiv
AI²-Active Safety: AI-enabled Interaction-aware Active Safety Analysis with Vehicle Dynamics

Wu, K., Li, Z., Li, S., Ye, X., Lord, D., Zhou, Y.

Under Review (2026)

[Preprint]

# active safety# vehicle dynamics# hypergraph

arXiv
AI²-Active Safety: AI-enabled Interaction-aware Active Safety Analysis with Vehicle Dynamics

Wu, K., Li, Z., Li, S., Ye, X., Lord, D., Zhou, Y.

Under Review 2026

[Preprint]

# active safety# vehicle dynamics# hypergraph

V2X-LLM: Enhancing V2X Integration and Understanding in Connected Vehicle CorridorsarXiv
V2X-LLM: Enhancing V2X Integration and Understanding in Connected Vehicle Corridors

Wu, K., Li, P., Zhou, Y., Gan, R., You, J., Cheng, Y., Zhu, J., Parker, S. T., Noyce, D., Ran, B., Tu, Z.

Under Review (2026)

[Preprint]

# LLM# V2X

arXiv
V2X-LLM: Enhancing V2X Integration and Understanding in Connected Vehicle Corridors

Wu, K., Li, P., Zhou, Y., Gan, R., You, J., Cheng, Y., Zhu, J., Parker, S. T., Noyce, D., Ran, B., Tu, Z.

Under Review 2026

[Preprint]

# LLM# V2X

Hypergraph-based Motion Generation with Multi-modal Interaction Relational ReasoningTR-C
Hypergraph-based Motion Generation with Multi-modal Interaction Relational Reasoning

Wu, K., Zhou, Y., Shi, H., Lord, D., Ran, B., Ye, X.

Transportation Research Part C: Emerging Technologies (2025)

[Paper]

# hypergraph# interaction# multi-modality

TR-C
Hypergraph-based Motion Generation with Multi-modal Interaction Relational Reasoning

Wu, K., Zhou, Y., Shi, H., Lord, D., Ran, B., Ye, X.

Transportation Research Part C: Emerging Technologies 2025

[Paper]

# hypergraph# interaction# multi-modality

A Digital Twin Framework for Physical-Virtual Integration in Enhancing Connected Vehicle Corridor Data PipelineIEEE T-ITS
A Digital Twin Framework for Physical-Virtual Integration in Enhancing Connected Vehicle Corridor Data Pipeline

Wu, K., Li, P., Cheng, Y., Parker, S. T., Noyce, D.,, Ran, B., Ye, X.

IEEE Transactions on Intelligent Transportation Systems (2025)

[Paper]

# digital twin# V2X# connected vehicle corridor

IEEE T-ITS
A Digital Twin Framework for Physical-Virtual Integration in Enhancing Connected Vehicle Corridor Data Pipeline

Wu, K., Li, P., Cheng, Y., Parker, S. T., Noyce, D.,, Ran, B., Ye, X.

IEEE Transactions on Intelligent Transportation Systems 2025

[Paper]

# digital twin# V2X# connected vehicle corridor

Graph-Based Interaction-Aware Multimodal 2D Vehicle Trajectory Prediction Using Diffusion Graph Convolutional NetworksIEEE T-IV
Graph-Based Interaction-Aware Multimodal 2D Vehicle Trajectory Prediction Using Diffusion Graph Convolutional Networks

Wu, K., Zhou, Y., Shi, H., Li, X., Ran, B.

IEEE Transactions on Intelligent Vehicles (2024)

[Paper]

# GNN# interaction# multi-modality

IEEE T-IV
Graph-Based Interaction-Aware Multimodal 2D Vehicle Trajectory Prediction Using Diffusion Graph Convolutional Networks

Wu, K., Zhou, Y., Shi, H., Li, X., Ran, B.

IEEE Transactions on Intelligent Vehicles 2024

[Paper]

# GNN# interaction# multi-modality

All publications