Democritus University of Thrace (DUTH)
Modeling the safety and efficiency performance of automated vehicles using fuzzy logic
Abstract
dc:descriptionIn recent years, there is a rapid evolution of automation and vehicle connectivity technologies. There are industry estimates that fully automated driving will be a reality in the coming years. Even for low levels of automation, these systems can have a significant impact on the state of road transport networks. Road safety is an important factor, as several accidents are due to human error, which can be avoided thanks to automated driving systems. In addition, traffic problems are very common in today's road networks, because of the growing demand for transportation and the inability of networks to provide adequate levels of service. Especially for urban networks, the effects of congestion are significant both in terms of fuel consumption and environmental impact, as well as in terms of the time that the average person loses in such a situation. Increasing the capacity of these networks can be very costly or even impossible with the current infrastructure. However, this can change with the advent of automation and vehicle connectivity. The driving behavior of automated vehicles can be designed to be more efficient, safer, and to make better use of network capacity. Developing systems that can be as secure and efficient as an experienced driver involve intricate problems and are by no means easy. Fuzzy logic can have several applications in these issues. The main research questions are: What is the current state of automated driving technology? What impacts can these technologies have on road networks? How can we have systems of guaranteed safety that do not lead to aggravation of traffic problems? Based on the results of the research of the above questions, we propose applications of fuzzy logic: Quantify the level of safety of a system of two vehicles car-following. Design a fuzzy automated longitudinal motion control system of a vehicle, which is safe and contributes to more efficient traffic flow. Develop fuzzy logic tools for more efficient vehicle routing. Simulation experiments were performed to assess the impact on road networks. Data were used to model the road network of the Antwerp ring road, a city in Belgium that is home to one of Europe's largest ports, resulting in major traffic problems. Major findings include that very conservative AV behavior can severely deteriorate the traffic condition, and connectivity may be necessary to improve congestion. Moreover, even for CAVs, if the demand increases, the potential benefits may not be realized, especially regarding environmental impacts. The behavior and capabilities of the existing systems were also analyzed. Using trajectory data from a real-world experiment, some important characteristics and capabilities of the systems are investigated. They are shown to be string unstable, increasing the magnitude of traffic shockwaves, which can be detrimental to traffic networks. In the context of the doctoral dissertation, mathematical models for recognizing unsafe driving behavior were researched in the literature. Their shortcomings in analyzing the level of safety of advanced technology systems were identified. Based on these, Surrogate Safety Metrics (SSMs) are developed based on definitions of fuzzy logic. The usefulness of the proposed SSMs and their advantages over traditional indicators have been investigated both by simulation experiments and on experimental data. The next step in the research is to investigate the usefulness of the proposed fuzzy SSMs for vehicle control. A fuzzy inference system was designed to identify potential hazards and react appropriately, depending on the level of danger. The proposed controllers are shown to be string stable and keep such distances that would facilitate the traffic flow.Another observation is concerning traffic demand. For increased demand, even if the vehicles form efficient platoons, congestion problems may worsen. Thus, prospects for optimal distribution of traffic demand may prove necessary. The problem of routing one or more vehicles often requires solving the well-known Traveling Salesman Problem (TSP). TSP is an NP-hard (Non-deterministic polynomial time hard) problem. Heuristic algorithms are often used to solve it. A fuzzy approach to compare the cost of each edge with all alternatives is developed, quantifying the suitability of each individual choice to be made. This approach is presented to improve the quality of solutions for a number of heuristic algorithms.
Degree
thesis:*- Grantor dc:publisher
- Democritus University of Thrace (DUTH)
- Year dc:date
- 2021
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Mattas, Konstantinos
- Μάττας, Κωνσταντίνος
Subjects
dc:subject × 12Rights
dc:rights- Statement dc:rights
-
- BY_4_0
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/49287