A Rule-Defined Adaptive MPC Based Motion Planner for Autonomous Driving Applications

In autonomous driving systems, motion planning to reach a given destination while avoiding obstacles becomes a task entirely managed by the on-board unit. In this work, we present a rule-defined motion planning algorithm for autonomous driving applications based on an adaptive Model Predictive Controller (MPC) framework. The motion planning task is first formulated as an Optimal Control Problem (OCP) subject to time-varying Control Barrier Function (CBF) constraints. It is then integrated within an MPC framework with adaptive weights settings, enabling the algorithm to dynamically adjust the MPC weights according to the rule-defined driving scenarios. The developed motion planner generates optimized trajectories for a high-fidelity Autonomous Vehicle (AV) model within IPG CarMaker software. Simulations performed showed that the developed motion planner adeptly facilitates successful overtaking, following, and stopping of the AV behind the Obstacle Vehicle (OV) based on rule-defined scenarios perceived by the AV.