Research Topics

My primary research interests lie in the intersection of control theory and machine learning towards applications to connected and autonomous vehicles (CAVs), nonlinear and complex systems. Some past/ongoing projects are summarized below.

1. NEXTCAR: Integrated Power and Thermal Management for Connected and Automated Vehicles (iPTM-CAV) Through Real-time Adaptation and Optimization

Check the related publications HERE

Future vehicles are expected to be able to exploit increasingly the connected and automated driving environment for efficient and safe driving. Additionally, connectivity and autonomous driving technologies open up new dimensions for control and optimization of vehicle and powertrain systems. While extensive studies have been carried out on fuel economy optimization for electrified vehicles, the implications of the connected and automated vehicles (CAVs) operation on integrated power and thermal management (TM) have not been fully explored. We focus on developing predictive control strategies enabled by advanced traffic modeling and CAV technologies to deliver power and heating (or cooling) for vehicle systems while achieve fuel saving. The detailed focuses of this research includes the following:

• Control-oriented prediction for i-PTM and adaption for improved prediction accuracy;

• Bench-mark development for i-PTM to evaluate the ceiling of the energy saving;

• Real-time climate control for CAVs in congested traffic.

2. Traffic and driver modeling by using BIG DATA Part 1: Evaluation of CAV energy efficiency in mixed traffic with human controlled vehicles

Check the related publications HERE

The CAV market penetration rate plays an important role in dynamic traffic. The current state of technology (low connectivity and low automation) mainly utilizes infrastructure based data to model traffic flow. As the penetration rate of CAV increases, more vehicles will become observable and controllable, so that current traffic models can be improved based on new sources of data. We focus the impact of mixed traffic flow on energy saving strategies under different penetration rates of CAVs. Scenarios such as merging, passing low speed vehicles, and cut-in that commonly happen in real traffic need to be studied with naturalistic driving data and considered in the control algorithms for better CAV trajectory planning.

3. Traffic and driver modeling by using BIG DATA Part 2: Deep learning-based vehicle velocity prediction and evaluation of energy saving in HEVs

Check the related publications HERE

The advanced transportation analytics and traffic environment information enable the characterization and quantification of the prediction for CAV operation, which will serve as critical inputs for CAV control strategies. Prediction of speed profile is one of the primary premises for CAV technology improvements to traffic system efficiency, drivability and safety. We focus on short-term prediction of vehicle velocity over 1~10 second and the long-term prediction of traffic flow over 20~30 minutes by using machine learning methods.