On Cooperative Control of Automated Driving Systems from a Stability and Safety Perspective

Doktorsavhandling, 2014

Over the last few decades, congested traffic network have become a serious problem in many countries. Congestions result in time losses, increase of fuel consumption and also raise the risk of accidents. Intelligent transportation systems may contribute to mitigate such problems. Advancement together with the reduction in cost of embedded computing, on-board vehicle sensors and wireless communication paved the way for introduction of automated driving systems. Vehicle platooning is an example of an automated driving systems which can be implemented to improve the traffic situation. To enable vehicle platooning with short inter-vehicles distances a control strategy is required that can guarantee passenger safety, comfort and stability of the platoon, so called string stability. While string stability is naturally defined in the frequency domain, stating safety and comfort requirements and vehicle limitations is more convenient as time domain specifications. Hence, fulfilling all the requirements and specifications simultaneously is not a trivial task. This thesis deals with the development of distributed model-based control strategies for a vehicle platoon. The aim of the control strategy is to enable platooning with a short inter-vehicle distance while fulfilling string stability criterion and maintaining safety and comfort. To achieve this, two approaches are proposed, i ) translating string stability criterion into time domain requirement and ii ) combining frequency domain control design techniques with Model Predictive control framework into a single control problem. Particular attention is given to ease the proposed methods for real time implementations. The control design is decoupled into longitudinal and lateral motion control and the methods presented can guarantee string stability and constraint satisfaction in both motion directions. Furthermore, a safety verification method based on reachability analysis technique and invariant set theory is proposed for safety analysis of automated driving systems for a given controller. The findings in this thesis are verified through simulations and field experiments.