Abstract
With the advancement of autonomous driving technologies, automated container terminals (ACTs) are transitioning toward mixed-traffic operations involving unmanned internal container trucks (ITs) and manned external container trucks (ETs). However, the complex interactions in mixed traffic present frequent conflicts and challenges for system performance evaluation. To address this, this study focuses on mixed-traffic ACTs and establishes a comprehensive performance evaluation model based on multi-agent simulation. First, three lane configurations are defined, including segregated unidirectional, mixed unidirectional, and mixed bidirectional. Then, a path interaction point (PIP) modeling method is proposed under traffic conflicts to characterize lane-level motion behaviors and conflict relationships represented by a conflict matrix, based on which an IT access strategy is developed. Second, a multi-level performance evaluation model for ITs is established, and a 3D multi-agent simulation model is developed to support performance evaluation under different lane configurations. Finally, the three lane configurations are evaluated through simulation at both individual and system levels under a baseline scenario, followed by a sensitivity analysis across varying IT–ET task ratios to assess adaptability. Simulation results indicate that the mixed bidirectional lane configuration achieves superior overall performance and robustness.