About Me
Hi, I'm Haopeng Deng, a graduate of Guangzhou Maritime University who recently completed summer research at MIT. I'm interested in how local interactions give rise to collective behavior and how a few targeted interventions can steer the system's evolution. My research explores traffic dynamics at both the flow and network scales, particularly in mixed-autonomy settings. I'm actively seeking PhD opportunities for Fall 2027.
Interests & Skills
Research Interests
Core Skills
Publications
Conference Papers
- Deng, H., He, S., & Wang, D. (2026). FLAT: Smoothing the Rugged Landscape for Learnable, Sample-Efficient Traffic Calibration. Submitted to The 38th IEEE Intelligent Vehicles Symposium (IV). [arXiv]
- Deng, H., Zheng, F., & Xia, X. (2026). SpinFlow: A Physics-Informed Spin Field Framework for Traffic Phase Inference and Transition Detection. The 29th IEEE International Conference on Intelligent Transportation Systems (ITSC). (Oral Presenter). [arXiv]
- Deng, H., Zheng, F., Ma, K., et al. (2025). Adaptive Truncated Schatten Norm for Traffic Data Imputation with Complex Missing Patterns. The 25th COTA International Conference of Transportation Professionals (CICTP). (Best Poster Award). [DOI]
- Xia, X., Deng, H.*, Ye, Y. (2026). Fundamental Diagram Modeling and Simulation of Mixed Traffic Flow Considering Functional Degradation. The 26th CICTP Conference. *Corresponding Author.
Journal Articles
- Xia, X., Deng, H., Tang, J., et al. (2025). Dynamic Traffic Assignment under Mixed Traffic: Modeling, Evolution, and Solution Approaches. Journal of Transport Science and Engineering. (Cover Paper, Oral Presentation at AIAT 2025). [DOI] [AIAT]
- Wang, Y., Li, L., Deng, H., et al. (2026). Sparse Stochastic Optimal Control under Control-Dependent Diffusion and Time-Varying Dynamics. Journal of Dynamical and Control Systems. [DOI]
- Deng, H., Tang, J., Zhan, L., et al. (2026). Research on Traffic Governance Optimization in Kaiping City Based on a GRA–FCE Integrated Framework. Journal of Transportation Engineering. [DOI]
- Deng, H., Ma, K., Zhan, L., et al. (2026). Structural Equation Modeling of the Relationships among Traffic Safety KAP in Small-Medium Cities. Journal of Transportation Engineering. [DOI]
Research Experiences
MIT–UF–NEU Joint Summer Research Camp
I worked with Ryan Hardesty Lewis in Prof. Cathy Wu's lab on inference-time behavior shaping for pretrained autoregressive driving models. I developed a framework that repeatedly selects the model's native short-horizon actions to steer a CAV and built a closed-loop platform integrating SMART/CAT-K, MetaDrive, and reactive IDM/MOBIL traffic. Using matched counterfactual rollouts and execution audits, I established repeatable CAV control across four traffic setups. I also traced traffic responses across six geometry–seed cases, linking the CAV's motion to neighboring vehicles' observations, applied controls, and speeds.
Guangdong Provincial Special Fund for Science and Technology Innovation Strategy
I led a project on cooperative traffic assignment, investigating the user equilibrium–system optimum continuum in mixed traffic. I introduced a two-dimensional framework based on automation and connectivity levels to characterize vehicle heterogeneity beyond CAV penetration alone. I formulated a two-class variational-inequality model separating cooperative vehicle share from cooperation level and explored adaptive mirror–projection methods for equilibrium computation. Simulations on a shared-bottleneck network showed that cooperation could reduce total travel time while increasing it for some origin–destination pairs. I also authored a systematic review of dynamic traffic assignment in mixed traffic, synthesizing over 120 sources on equilibrium evolution, solution algorithms, and research trends. I coordinated team tasks and mentored junior researchers.
Guangdong Provincial College Student Innovation and Entrepreneurship Training Program
As a core member, I contributed to developing an intelligent system for recognizing urban traffic equilibrium states, integrating data from roadside sensors, UAVs, and probe vehicles. My team aimed to construct high-dimensional traffic flow tensors and explore data loss structures—including element-wise, fiber-wise, and mixed loss—to improve data recovery algorithms. I developed an adaptive truncated Schatten norm-based model (LRTC-ATSN) for recovering traffic data with complex missing patterns, improving MAPE by 10.6% and RMSE by 6.1%. I also proposed SpinFlow, a physics-informed model driven by a latent spin field to capture competitive equilibrium among traffic phases, combined with an Expectation-Maximization algorithm to infer synchronized flow emergence from aerial trajectory data.
High-Quality Development Project for Hundred Counties, Thousand Towns, Myriad Villages
I led a team investigating traffic safety and planning in Kaiping City, working with the Kaiping Municipal Transportation Bureau to analyze accident and violation data and identify crash hotspots and hazard types. My team conducted on-site surveys, interviews, and drone photography to gather data on traffic flow, vehicle management, pedestrian movement, and infrastructure. I applied SPSS and LLM tools to analyze the survey and interview data, identifying recurring issues and proposing solutions that were later adopted by the Municipal Transportation Bureau. I also authored two papers from this project: one using Structural Equation Modeling to examine the relationship between traffic safety behavior and attitudes, and another assessing Kaiping's traffic governance through Grey Relational Analysis and Fuzzy Evaluation.
Academic Visits
MIT-UF-NEU Joint Summer Research Camp
Selected as one of 41 participants, I joined a research community spanning urban science, transportation engineering, and AI. Exchanges with faculty, mentors, and peers deepened my interest in learning and control for autonomous systems, especially how agents make decisions when others adapt to their actions.
Smart Sustainable Cities and Transportation Winter School
Selected as one of 67 participants, I completed 11 modules on data-driven modeling, digital twins, and autonomous driving. It helped me see transportation problems not just as technical optimization, but as societal systems, where algorithmic design needs to account for real-world uncertainty and human factors.
Awards & Honors
Off the Clock
I enjoy wandering around cities and taking photos of whatever catches my eye. At home, I'm often reading and writing, making pour-over coffee, or trying something new in the kitchen.