A hybrid hierarchical rally driver model for autonomous vehicle agile maneuvering on loose surfaces

This paper presents a novel Hybrid Hierarchical Autonomous system for improving vehicle safety based on agile maneuvering and drift control on loose surfaces. Standard Electronic Stability Control Systems provide stability by limiting the vehicle body slip, thus reducing the capability of the vehicle to generate lateral acceleration and follow road segments and paths with high curvature on loose surfaces. The proposed system overcomes this shortcoming. Furthermore, it is the first time where a solution for arbitrary road geometries is proposed. The system described in this work consists of three layers. The first layer selects the driver model. The second layer selects the path to be followed and the maneuver type using a Proportional controller and motion planning strategies. The third layer coordinates the steering and driving functions of the vehicle to perform the maneuver, where a Gain-Scheduled Linear Quadratic Regulator is employed to achieve drift control. The hybrid system is implemented in Matlab/Simulink R and tested in two scenarios: First, a Rally-like stage formed by a combination of clothoid and arc segments is used to study the drift-path-following capabilities of the system, and lastly, a lateral collision case is proposed to evaluate the suitability of the system as an ADAS Co-Pilot system for lateral collision avoidance.