CAV Efficiency at Stop-Controlled Intersections

Overview

A graduate MATLAB simulation project evaluating how Connected and Automated Vehicles (CAVs) could improve operations at stop-controlled intersections compared with human-driven vehicles (HDVs). The work developed a custom simulation framework that contrasts HDV and CAV operating parameters, layers a safety-check conflict-resolution algorithm on top of the scheduling logic, and quantifies throughput and delay performance through both a detailed single-run case and a large-scale randomized multi-run experiment.

Tools Used

MATLAB · Discrete-Event Simulation · Headway & Switching-Time Modeling · Block-Priority Scheduling · Safety-Check Conflict Resolution · Monte Carlo / Multi-Run Experimentation

My Role

Designed and implemented the simulation framework end-to-end: parameterized HDV vs. CAV behavior, built the scheduling and safety-check logic, ran the detailed single-run case, set up and executed the 50,000 randomized multi-run experiment, and interpreted the comparative results.

Method

The framework modeled vehicles arriving at a stop-controlled intersection with separate parameter sets for HDV and CAV behavior — including following headways, switching times between conflicting movements, and reaction characteristics. A block-priority scheduler assigned the next conflict-free movement, and a safety-check algorithm sat on top of the scheduler to resolve conflicts before they could materialize. Performance was first explored through a detailed single-run case to validate logic and surface intersection-level dynamics, then evaluated at scale through 50,000 randomized trials sweeping over intersection complexity (number of approaches, demand mix, and arrival randomness) so that the comparison between HDVs and CAVs would generalize beyond a single configuration.

Key Outcomes

• Built a working MATLAB simulation framework that supports both HDV and CAV parameter sets at stop-controlled intersections.
• Added a safety-check conflict-resolution layer on top of block-priority scheduling so that CAV efficiency gains are not produced by unsafe assumptions.
• Used a detailed single-run case to validate the model and illustrate intersection-level behavior.
• Ran a 50,000-trial randomized multi-run experiment across varying intersection complexity to test how robust the CAV gains are.
• Showed that CAVs reduced delay and increased throughput relative to HDVs, with the gap growing as intersection complexity and demand increased.

Limitations

The framework abstracts vehicle behavior to scheduling-level parameters rather than full vehicle dynamics or sensor models, so it captures coordination benefits more cleanly than physical control limits. The safety check is rule-based and assumes reliable communication between CAVs — degraded connectivity, mixed-fleet edge cases, and pedestrian/bicycle interactions were not modeled. Results are bounded to stop-controlled intersection geometries and do not directly translate to signalized or uncontrolled cases.