Design and evaluation of CCA (Cooperative Collision Avoidance) applications for vehicular ad-hoc networks

Abstract

[SPA] El tema central de la Tesis ha versado sobre el diseño y evaluación de aplicaciones para la reducción de la probabilidad de colisión en carretera mediante el uso de conectividad inalámbrica entre vehículos, particularmente en un escenario específico del tráfico rodado: presencia de un obstáculo en la dirección de tránsito que bloquea el paso. Dos enfoques han sido tomados en consideración: utilización de mecanismos de anticipación cooperativa vehículo a vehículo para evadir colisiones mediante frenada, y empleo de esquemas de maniobras de evasión cooperativa en circunstancias donde existe suficiente espacio en la carretera para reorientar las trayectorias y evitar el choque. Se ha hecho uso de herramientas de simulación de redes y dinámica vehicular, y de la teoría matemática de la optimización y de los procesos estocásticos para modelar estos escenarios. Los resultados demuestran que el uso de comunicaciones, junto con sistemas avanzados de inteligencia artificial permitirá en un futuro garantizar cotas de seguridad en carretera nunca antes vistas, incluso en situaciones de riesgo extremo que podrían ser detectadas por uno o más vehículos con tiempos muy cortos de reacción.

[ENG] New emerging technologies in vehicular traffic are aimed primarily at improving safety and driving comfort for passengers, by paying special attention to the gradual evermore automation of all aspects of the driving task. In this regard, a promising research perspective considered by the Academia and the Industry is to use communications to build a complex interoperable vehicular network that would serve as a means to provide autonomous robotic-guided vehicles with additional status information that might not be collected from sensors on board. With properly configured processing schemes, this additional stream of information can be used to help vehicles anticipate and react conveniently to potentially risky situations that might cause an accident if not previously considered. Particularly, in this Thesis we use these premises to propose and evaluate collision avoidance policies under two specific fashions: i) Design and evaluation of a Cooperative chain Collision Avoidance (CcCA)1 strategy to reduce the impact of multiple rear-end collisions in a platoon of vehicles when evasive maneuvering is not possible, and ii) Analysis and optimization of different strategies for Cooperative Collision Avoidance (CCA) by evasive maneuvering. The CcCA application allows us to study how communication protocols, both by one-hop transmissions as well as by relaying (multi-hop) schemes, can help reduce the number of accidents, or at least minimize their impact, in cases where vehicles cannot execute sudden maneuvers to skip cars ahead, but only brake. Simulations are validated by using an advanced stochastic model which rigorously describes the behavior of vehicles in this type of situations. Among other aspects, results show that real implementations of CcCA must take into account with special relevance those vehicles that might be humanly driven, and guarantee that during the transition stage (until a complete penetration of the technology is achieved) safety is preserved enough. Regarding CCA for evasive maneuvering, we provide an exhaustive optimization analysis for the calculation of optimum trajectories in cases where vehicles at high speeds are at risk of colliding with one or more obstacles appearing ahead. By reorienting trajectories through the lateral free spaces that might exist between the obstacles and the crash barriers (if the specific scenario allows it), vehicles can avoid crashing and simultaneously improve driving comfort even under such unpredictable circumstances. On the whole, despite much further effort is still required on these matters, results in this Work show that communications can help autonomous vehicles to make decisions in a cooperative fashion that will not only assist individuals to follow the best riding strategy, but also the traffic system as a whole to evolve according to the best possible behavior in terms of safety and comfort.