Autonomous Highway Path-Planning: Leveraging Localized Quintic Spline Trajectories

This paper presents a simple and effective Localized Spline-based Path-Planning (LSPP) protocol for autonomous cars navigating highways. The primary goal is to enable the autonomous ego-car to maneuver smoothly, avoid obstacles (such as moving cars), prioritize comfort, and adhere to lane boundaries and other road constraints. The LSPP algorithm generates an optimized trajectory using spline functions and a speed profile, which the autonomous vehicle follows using its actuators. The egocar's sensor fusion module detects surrounding moving cars and other obstacles, providing accurate localization input to the LSPP module. This module then sends the generated safe trajectory to the model-predictive-control (MPC) track-follower for precise tracking, a comfortable ride, and efficient travel time. Extensive simulations on complex highways with various driving scenarios were conducted to assess the LSPP's performance at different speeds. The results and analysis demonstrate that the proposed LSPP, with its tuning technique, is highly effective and promising.