Predictive Control for Waypoint Path Settling Maneuver with Robustness to Initial Vehicle Pose

Abstract

Recently, interest in autonomous vehicles has increased. Among autonomous driving control technologies, waypoint tracking is a method of driving along a global route created by a global navigation satellite system and is essential from the perspective of lateral motion control. In waypoint tracking control, if an ego vehicle's position and/or direction angle has a sizeable lateral offset with the planned route, the control input may diverge or cannot be computed. There may also be an initial situation when the ego vehicle starts to track the waypoints opposite to the desired direction. In this study, we propose a way for settling to the global route robustly with respect to the initial pose via model predictive control. We derived an integrated motion model of longitudinal and lateral direction. For validation of the proposed method, the experiment was conducted through computational simulation. We confirmed that the vehicle smoothly entered the reference route regardless of the initial position and heading of the ego vehicle.