A Comparative Conflict Resolution Dataset Derived from Argoverse-2: Scenarios with vs. without Autonomous Vehicles

This research tackles the important problem of how vehicles resolve conflicts at unsignalized city intersections. It compares situations where autonomous vehicles (AVs) are present with those that involve only human‑driven cars. The study uses the open‑source Argoverse‑2 motion‑forecasting dataset. Interaction scenarios are selected by applying simple filtering rules that rely on surrogate safety measures such as: – post‑encroachment time (PET) – minimum spatial distance between vehicles – points where trajectories cross – changes in speed The raw dataset has quality issues—position and speed data sometimes do not match. To fix this, a robust processing pipeline was built. It uses kinematic correction and smoothing algorithms to rebuild very accurate trajectories, giving

consistent values for position, speed, acceleration, and heading direction. After processing, the refined dataset contains: – more than 5,000 conflict scenarios that include an AV – more than 16,000 conflict scenarios with only human‑driven vehicles This clean, structured data set serves as a solid base for: – comparing traffic safety in mixed and human‑only traffic – modeling driver behavior in mixed traffic streams – testing motion‑planning algorithms for autonomous driving in complex urban settings With this framework, researchers can closely examine how autonomous decision‑making changes the way conflicts are resolved at intersections.

Message: Hi FE, I need help with “A Comparative Conflict Resolution Dataset Derived from Argoverse-2: Scenarios with vs. without Autonomous Vehicles” in “Civil Engineering”