Research
Physics-Informed Digital Twin-Enhanced Reinforcement Learning in Mixed-Fleet Mobility-on-Demand Systems
Advisor: Dr. Kaidi Yang in the Dept. of Civil & Environmental Engineering at National University of Singapore
Developed a model-free reinforcement learning method for vehicle rebalancing in mixed-autonomy mobility-on-demand systems that integrated a physics-informed digital twin to improve sample efficiency. This research has resulted in a research paper under review in Transportation Research Part C
Reliable Real-Time Evacuation Using Uncertainty-Informed Model Predictive Control
Advisor: Dr. Kaidi Yang in the Dept. of Civil & Environmental Engineering at National University of Singapore
Developing an online vehicle rerouting algorithm for real-time evacuation under adverse weather conditions based on uncertainty-informed model predictive control that leverages conform prediction to quantify uncertainty.
Congestion-aware Reinforcement Learning in Autonomous Mobility-on-Demand Systems
Advisor: Dr. Kaidi Yang in the Dept. of Civil & Environmental Engineering at National University of Singapore
Developing a congestion-aware reinforcement learning-based algorithm for vehicle rebalancing in autonomous mobility-on-demand systems where congestion is reflected by a link transmission model.
Privacy-Concerned Trajectory Generation Using Differential Privacy Generative Adversarial Network
Advisor: Dr. Kaidi Yang in the Dept. of Civil & Environmental Engineering at National University of Singapore
Developed a novel presentation of trajectories which are further clustered based on distances and generated using differential privacy generative adversarial network
Risk Evaluation and Protection of Overtaking based on Simulated Driving Experiment
Advisor: Prof. Yongjun Shen in the Dept. of Transportation Engineering at Southeast University
Conducted state preference survey to evaluate human drivers’ subjective aggressiveness during overtaking and developed regression models to evaluate human drivers’ objective aggressiveness during overtaking based on data collected from simulated driving experiments. This program has been selected as the Jiangsu Provincial-level undergraduate research program
Antenna Array Optimization based on Heuristic Algorithms
Advisor: Dr. ZhongJin Jiang in the Dept. of Information Science and Engineering at Southeast University
Developed a particle swarm optimization algorithm to efficiently optimize antenna array performance where fitness function is specially designed and various simulation environments (i.e., diverse signals) to evaluate the algorithm effect.