Une approche dynamique pour la gestion de feux de circulation avec les voitures connectées
|Abstract:||The most practical and economical solution to reduce congestion in the cities is to improve traffic control systems, especially traffic lights signals. These systems have significant impacts on waiting times, accident risks and unnecessary fuel consumption. The majority of these systems are however static, that is to say that the programming is fixed or pre-timed. They are therefore not receptive to demand. The two main objectives of this thesis were to reduce the complexity for the management of traffic lights in real time and to propose a model to collect the data necessary to apply our approach in real life. The approach proposed to achieve the first goal, is inspired by the work of a human circulation officer. It consists of reducing stops and encouraging group travel of vehicles such as traffic officers do in real life. To achieve this, the traffic density was measured continuously, and the traffic light sequences were modified accordingly. To test our approach, a busy sector of Quebec City was identified and a simulation with a microscopic simulator was performed. Waiting times have been reduced significantly. Our second goal was to transfer this solution in real life. However, a fundamental challenge was to measure traffic density at every moment as request in our approach. This task is expensive and requires installing video devices or other sensors scattered over the network. A new service architecture model has therefore been developed to work around this problem and it relies on connected cars technology. This technology allows to obtain the vehicle position every 0.1 second on a road network and it was the necessary ingredient to apply our approach in real life. From this value, it was then possible to calculate the traffic density and apply the proposed solution. Thus, data collection, transmission to a traffic management center, processing and application of the solution could be done instantly at an economical cost. This thesis therefore shows that it is possible to improve the performance of current traffic light systems by applying rules based on common sense and to apply a practical and economical implementation method to transpose them into real life. Furthermore, a limitation of our work is that the simulation process is an overview of the reality and it is difficult to predict the results of the experiment in other contexts. It would therefore be advantageous to continue research in other environments. In addition, the technology of connected cars is not yet deployed in Canada. However, it is hoped that this technology is about to emerge with current massive investment by vehicle manufacturers and deployment of 5G. A test bed would be the next step to test the solution proposed in situ.|
|Document Type:||Thèse de doctorat|
|Open Access Date:||20 December 2021|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.