A Decentralized Approach for Connected and Autonomous Vehicles Traffic Negotiation and Conflict Resolution

• Ano de defesa: 2023
• Autor(a) principal: Santos, Tiago Cesar dos
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Collective probabilities; Connected autonomous vehicles; Decentralized negotiation; Multi- objective; Multi- objetivo; Negociação descentralizada; Probabilidades coletivas; Veículos autônomos conectados
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/55/55134/tde-04072023-134313/

Population growth and fast urban expansion in the last century generated a metropolitan agglom- eration that caused numerous social and economics impacts. The high demand for mobility has resulted in a growing number of vehicles on the roads, leading to longer travel times, which has caused traffic congestion, increased accident rates, and increased pollution. Intelligent Transportation Systems (ITS) have the potential to revolutionize the transportation and urban mobility to reduce traffic accidents, mitigate congestion and traffic jams. The communication between vehicles allows negotiation and cooperation that can be used in various contexts such as crossing intersections, on-ramp merge and lane change. Resolving conflicts between vehicles is a challenging task due to the higher density of vehicles and limited maneuverability, often limited to only speed changes. A significant number of studies focus on solutions rely on centralized methods that have a heavy dependency on the infrastructure. Additionally, there is limited research on decentralized negotiation methods that take into account multi-objective systems where each vehicle has its own private cost function and also there is a lack of work in generalized solutions that tackles different dynamic scenarios with traffic flow using the same approach. This thesis proposes a method for resolving conflicts between vehicles that have the ability to communicate in a decentralized manner that can be used in multiple dynamic scenarios, such as: crossing intersections, on-ramp merging and changing lanes. We also considered vehicles with different characteristics, dimensions and private cost functions. The proposed approach aims to improve the flow of vehicles and reduce travel time that can be translated into a reduction in the emission of pollutants. To evaluate our proposal, we use a simulation tool to create various test scenarios and compare our results with existing methods using metrics commonly used in the literature. In addition, it was also necessary to adapt the simulation tool to include features such as trajectory calculation, collision detection and vehicle management in relation to the decision-making process. In every scenario, significant improvements were achieved in terms of vehicle flow, resulting in a decrease in travel time and an increase in average speed when compared to the benchmarks used.