Safety Assessment for Precautionary Collision Avoidance - A Collision Avoidance System Based on Safety-Set

Abstract

Abstract This thesis presents the implementation of a precautionary collision avoidance system, centered on obtaining a safety set and employing Model Predictive Control (MPC) based on the safety set. The safety set, which defines the boundaries within which the vehicle can safely operate, is derived through optimization problems or analytical calculations. By applying the approaches, the system proactively anticipates and mitigates potential collision risks, such as hidden pedestrians emerging from blind spots. Two critical scenarios are examined: the hidden pedestrian scenario, where the vehicle must react to unforeseen obstacles, and the lane change scenario, which requires safe merging while considering nearby traffic. The integration of the precautionary safety-set with MPC allows the system to maintain a high success rate in collision avoidance while ensuring smooth and efficient vehicle operation. Experimental results confirm the system’s reliability and demonstrate its ability to maintain a high success rate in safely navigating complex scenarios, thus highlighting its potential to significantly enhance the safety and performance of autonomous driving systems.