Optimizing Vehicle Trajectories at a Signalized Intersection in Mixed Traffic
Cheng Peng, Jiaping Wang, Pengchao Liu, Zhen Wang, Xiangmo Zhao
AI summary
Problem
Existing CAV trajectory optimization in mixed traffic overlooks human driver variability and relies heavily on simulations, leaving a gap in real-world adaptive control and platoon guidance strategies.
Approach
The method uses real-world car-following data to train an improved Informer model for predicting CAV arrival states, then applies Bayesian optimization to tune MPC parameters by learning human driving characteristics.
Key results
- Improved Informer model accurately predicts CAV arrival states at intersections
- Bayesian optimization tunes MPC parameters by learning human driving features
- Real-vehicle experiments validate smoothed speed profiles and enhanced platoon efficiency
- Method adapts to varying CAV platoon positions and dynamic signal phases
Why it matters
Enables safer, more efficient integration of autonomous vehicles into current road networks, guiding urban traffic management and AV development.
Abstract
With the advancement of connected and automated vehicles (CAVs), achieving accurate vehicle trajectory prediction and optimal control has become a critical challenge for improving the efficiency and safety of mixed traffic flow. However, due to the complex dynamic interactions between CAVs and human-driven vehicles (HVs) and the nonlinear nature of signal coordination, existing studies lack comprehensive consideration of CAV position adjustments within the platoon and their guidance effects on trailing HVs. We propose a data-driven method for CAV state prediction and trajectory optimization. Employing an application- specific improved Informer model, our method accurately predicts CAV arrival states at a signalized intersection in mixed traffic. Additionally, Bayesian optimization is utilized to achieve automated and rapid tuning of CAV model predictive control parameters through learning human driving characteristics. Experimental results demonstrate that our proposed method significantly enhances overall traffic efficiency and optimization when CAVs operate within mixed traffic, showing strong feasibility and adaptability.