Optimal vehicle dynamics control and state estimation for a low cost GNSS-based collision avoidance system

• Optimale Fahrdynamikregelung und Zustandsschätzung für ein Low-Cost GNSS-basiertes Kollisionsvermeidungssystem

Katriniok, Alexander; Abel, Dirk (Thesis advisor)

Als Ms. gedr.. - Düsseldorf : VDI-Verl. (2014)
Dissertation / PhD Thesis

In: Fortschritt-Berichte VDI / Reihe 8, Meß-, Steuerungs- und Regelungstechnik 1230
Page(s)/Article-Nr.: X, 198 S. : Ill., graph. Darst.

Zugl.: Aachen, Techn. Hochsch., Diss., 2013

Abstract

The context of this dissertation is the investigation and experimental evaluation of a global navigation satellite system (GNSS) based automotive collision avoidance system in the scope of the project Galileo above. While the application of collision avoidance comprises a variety of research topics, this thesis focuses on the issues of navigation, vehicle state estimation and vehicle guidance at the handling limits. The main aim of this thesis is to investigate and implement a concept for conducting autonomous evasion maneuvers based on a low-cost GNSS receiver (absolute horizontal position accuracy of about 4 m) and inertial sensors. In previous publications, centimeter-precision high-cost navigation systems (absolute horizontal position accuracy in the centimeter range) have commonly been employed for this purpose. To solve the navigation task, a navigation concept based on relative positioning is introduced that is appropriate for evasion maneuvers having high horizontal accelerations greater than 7 m/s^2. Vehicle state estimation is related to determining key drive dynamic vehicle states (especially the longitudinal and lateral velocity at the center of gravity) which can generally not be measured using series production sensors but are required for the considered control concept. The estimator is designed in such a way that these vehicle states can be determined appropriately even at the handling limits. Furthermore, the concept allows for being adaptive with respect to uncertainties in the tire-road contact. As the evasion path respectively trajectory is considered to be given over a finite time horizon and physical constraints like actuator limitations and the tire-road friction limit are taken into account, a model predictive control scheme is employed to guide the vehicle autonomously. Besides the discussion of the proposed algorithms, experimental results are presented and evaluated.