About Me

I’m a Ph.D. student at McGill University under the supervision of Prof. Lijun Sun. I’m currently also a visiting researcher at the Robotics Institute, Carnegie Mellon University (CMU), supervised by Prof. Changliu Liu. Previously, I was a visiting researcher with the Department of Mechanical Engineering, UC Berkeley from 2019 to 2020, supervised by Prof. Masayoshi Tomizuka; and was with the Department of Mechanical Engineering at CMU in 2018.

My research interests are Bayesian learning, macro/micro driving behavior analysis, and multi-agent interaction modeling in intelligent transportation systems. Specifically, I’m more interested in revealing the mechanisms of social interactions conveyed in microscopic human driving behaviors and investigating how the interactive behaviors impact the macroscopic traffic flow dynamics.

Please feel free to contact me if you are interested in collaborating with me. I’m also actively seeking undergraduates and master students who are interested in a summer/remote research internship.


  • New! 06/2023: I’m co-organizing the 1st International Workshop on Socially Interactive Autonomous Mobility (SIAM) at IV23’ in Anchorage, Alaska, USA. Welcome to join us on June 4th, 2023!
  • New! 02/2023: I presented our work “Bayesian Calibration of the IDM” on the 2023 Traffic Flow Theory and Characteristics (ACP50) general webinar series. [Flyer] [Recording]
  • New! 11/2022: The preprint of our recent work “Discovering dynamic patterns from spatiotemporal data with time-varying low-rank autoregression” is available on arXiv. The adapted datasets and Python implementation are available at https://github.com/xinychen/vars.
  • New! 11/2022: Our review paper on Social Interactions for Autonomous Driving is accepted by Foundations and Trends® in Robotics.


Featured Research

Social Interactions for Autonomous Driving: A Review and Perspective.

Wenshuo Wang, Letian Wang, Chengyuan Zhang, Changliu Liu, and Lijun Sun. Foundations and Trends in Robotics.

[Abstract] No human drives a car in a vacuum; she/he must negotiate with other road users to achieve their goals in social traffic scenes. A rational human driver can interact with other road users in a socially-compatible way through implicit communications to complete their driving tasks smoothly in interaction-intensive, safety-critical environments. This paper aims to review the existing approaches and theories to help understand and rethink the interactions among human drivers toward social autonomous driving. We take this survey to seek the answers to a series of fundamental questions: 1) What is social interaction in road traffic scenes? 2) How to measure and evaluate social interaction? 3) How to model and reveal the process of social interaction? 4) How do human drivers reach an implicit agreement and negotiate smoothly in social interaction? This paper reviews various approaches to modeling and learning the social interactions between human drivers, ranging from optimization theory, deep learning, and graphical models to social force theory and behavioral & cognitive science. We also highlight some new directions, critical challenges, and opening questions for future research.

Spatiotemporal Learning of Multivehicle Interaction Patterns in Lane-Change Scenarios

Chengyuan Zhang, Jiacheng Zhu, Wenshuo Wang, and Junqiang Xi. IEEE Transaction on Intelligent Transportation Systems.

[Abstract] Interpretation of common-yet-challenging interaction scenarios can benefit well-founded decisions for autonomous vehicles. Previous research achieved this using their prior knowledge of specific scenarios with predefined models, limiting their adaptive capabilities. This paper describes a Bayesian nonparametric approach that leverages continuous ( i.e., Gaussian processes) and discrete (i.e., Dirichlet processes) stochastic processes to reveal underlying interaction patterns of the ego vehicle with other nearby vehicles. Our model relaxes dependency on the number of surrounding vehicles by developing an acceleration-sensitive velocity field based on Gaussian processes. The experiment results demonstrate that the velocity field can represent the _spatial_ interactions between the ego vehicle and its surroundings. Then, a discrete Bayesian nonparametric model, integrating Dirichlet processes and hidden Markov models, is developed to learn the interaction patterns over the _temporal_ space by segmenting and clustering the sequential interaction data into interpretable granular patterns automatically. We then evaluate our approach in the highway lane-change scenarios using the highD dataset collected from real-world settings. Results demonstrate that our proposed Bayesian nonparametric approach provides an insight into the complicated lane-change interactions of the ego vehicle with multiple surrounding traffic participants based on the interpretable interaction patterns and their transition properties in temporal relationships. Our proposed approach sheds light on efficiently analyzing other kinds of multi-agent interactions, such as vehicle-pedestrian interactions.

Bayesian Calibration of the Intelligent Driver Model

Chengyuan Zhang and Lijun Sun. Available on arXiv.

[Abstract] Accurate calibration of car-following models is essential for investigating microscopic human driving behaviors. This work proposes a memory-augmented Bayesian calibration approach, which leverages the Bayesian inference and stochastic processes (i.e., Gaussian processes) to calibrate an unbiased car-following model while extracting the serial correlations of residual. This calibration approach is applied to the intelligent driver model (IDM) and develops a novel model named MA-IDM. To evaluate the effectiveness of the developed approach, three models with different hierarchies (i.e., pooled, hierarchical, and unpooled) are tested. Experiments demonstrate that the MA-IDM can estimate the noise level of unrelated errors by decoupling the serial correlation of residuals. Furthermore, a stochastic simulation method is also developed based on our Bayesian calibration approach, which can obtain unbiased posterior motion states and generate anthropomorphic driving behaviors. Simulation results indicate that the MA-IDM outperforms Bayesian IDM in simulation accuracy and uncertainty quantification. With this Bayesian approach, we can generate enormous but nonidentical driving behaviors by sampling from the posteriors, which can help develop a realistic traffic simulator.

Thanks & Fundings