Cooperative Connected Intelligent Vehicles and Infrastructure for Road Safety Applications

Abstract

Connected Vehicles (CV) and Autonomous Vehicles (AV) are two of the most promising modern vehicular technologies to tackle road safety challenges with advanced communications, sensor and Artificial Intelligence (AI) technologies. The primary motivation for this thesis is a study into the design and evaluation of connected and autonomous vehicles (CAVs) technologies for cooperative road safety applications.
As vehicular communication plays a critical role in CAV networks and road safety applications, this thesis designs, improves and evaluates a software-defined radio (SDR) based vehicle to everything (V2X) platform. It provides a scalable and more cost-effective solution for Dedicated Short-Range Communications (DSRC) and V2X research. Experiment results show the SDR-based DSRC can perform to the standards of commercial DSRC systems with high reliability of over 99%, low latency of 2.89 milliseconds and a communication range suitable for Vehicle to Vehicle (V2V) which is measured up to 250 metres. It maintained high performance whilst mobile and is proven capable of transmitting video data. This improved SDR-Based DSRC is an invaluable academic tool for the research community.

Furthermore, the V2X communication reliability and its impact on road safety are investigated. This is essential for modelling safety with communication capabilities. New algorithms for collision avoidance with V2V communication are designed and developed. V2V communication packet losses are measured and modelled. Its impact on stopping and safety distance is investigated. Simulation results show that V2V can effectively reduce vehicle reaction time, which decreases the stopping distance needed, and that the density of CV networks has a more significant impact on safety distance than the V2V communication distance. These investigations show significant impacts to safety when consecutive packet losses are incurred, leading to increased numbers of collisions.

Lastly, cooperative sensing among the CAVs and connected road infrastructure is investigated to improve road safety. As the sensors of AVs have inherent shortcomings, cooperation among CAVs and the infrastructure could help enhance the perception of the driving environment and make safe driving decisions. A novel Cooperative connected Road-infrastructure and Autonomous Vehicles (CRAV) framework is designed and developed with a scalable and flexible simulation approach. Various use cases with CRAV are studied to investigate the vulnerable road users (VRU) collision warning. A new approach to utilising the Responsibility-Sensitive Safety (RSS) model and CRAV is also designed to provide an appropriate safety distance to facilitate collision avoidance. Results show that with the support of Roadside Units (RSU), vehicles can be alerted to the presence of VRUs much earlier than when using local sensors, up to a maximum of 80 metres. RSS with CRAV integration can significantly reduce road collisions, which is highly significant for reducing road fatalities and the results show collisions are reduced with the incorporation of these new algorithms for speeds up to 31 metres per second.

To sum up, this thesis presents an improved SDR based DSRC communication tool for academic research into V2X, an extensive investigation and creation of new algorithms for stopping distance and a novel CRAV framework simulation that compiles many areas needed for V2X into one complete system. The approaches taken in this thesis show significantly improved safety for road users.

Future work will be an extension of the SDR-based DSRC in large scale field tests and an incorporated automatic gain control. In addition, CRAV will be further developed to consider an expanded amount of safety scenarios and an investigation into the impacts of CRAV on road capacity.

Date of Award	Jan 2022
Original language	English
Supervisor	Zuoyin Tang (Supervisor) & Stylianos Sygletos (Supervisor)