Automated Driving Control in Highway Scenarios Through a Two-Level Hierarchical Architecture

Abstract

We introduce an approach for automated driving in highway scenarios based on a two-level hierarchical architecture. The high-level consists of a path planner implemented through a Model Predictive Control algorithm that, using a simple kinematic model of the vehicle, effectively manages the relevant maneuvers of highway driving, such as lane keeping, lane change, velocity, and distance tracking, by means of suitable combinations of artificial potential field functions. Parameters of such functions are dynamically tuned according to the acquired scenario. A switching logic described by a finite state machine, based on acquired sensor data, selects the most appropriate maneuver to realize in the present driving scenario. At the low-level, the motion controller regulates the longitudinal and lateral dynamics through an original decentralized architecture to track the generated reference trajectory. Robustness issues in the presence of plant uncertainty are handled by $H_{\infty } $ synthesis. Extensive simulation tests show the effectiveness of the proposed approach.

Keywords

• Autonomous vehicles
• model predictive control
• robust control
• path planning