Rural Intersection Safety for Autonomous and Connected Vehicles (CTS)

Summary

Ensuring safety at intersections remains one of the most critical challenges in autonomous driving, given the complex interplay of vehicle trajectories, unpredictable behaviors of human drivers, and ambiguous right-of-way rules. This report investigates the use of Control Barrier Functions (CBFs) as a rigorous control strategy for guaranteeing collision-free behavior in such environments. Wireless information from the intersection on the trajectories of cross-traffic vehicles is assumed to be available. Through extensive testing in MATLAB and high-fidelity CARLA simulations, we demonstrate that CBF-based controllers can robustly enforce minimum distance constraints and adapt to dynamic cross-traffic conditions in real time.

The core technical contributions of this work include the formulation of both reactive and predictive CBF models, implemented as Quadratic Programs (QPs) that modify nominal acceleration commands to maintain safety. These simulations account for measurement noise, uncertainty in cross-traffic behavior, and realistic sensing constraints.

To complement the simulation efforts, a real-world proof-of-concept was conducted using the MnCAV research vehicle equipped with GNSS-based position control and a radar-powered intersection monitoring unit that transmitted simplified intersection status updates via XBee radio modules. While the CBF controller was not deployed onboard during these field tests, the experiments served to validate system behavior under simplified intersection conditions and highlight the feasibility of infrastructure-assisted intersection safety.

This layered validation approach from simulation to experimental deployment offers a promising roadmap for advancing safety centric autonomous vehicle technologies at intersections.