Autonomous Collision Avoidance based on Linear Time-Varying Model Predictive Control

This paper introduces an autonomous collision avoidance framework for vehicles, encompassing trajectory planning and path tracking. A 3D potential field identifies safe driving zones, considering road lanes, obstacles, and environmental factors. A quintic polynomial path ensures path continuity and smoothness. The controller treats path tracking as a linear time-varying model predictive control task. The front steering angle is meticulously calculated for effective steering. A hardware-in-the-loop system is structured, and tests across varying velocities and road surfaces demonstrate the framework's effectiveness and maneuverability.