Research on driverless cars is generally vehicle centric. The technological development achieved is impressive, and the potential and attractiveness of the concept are huge. However, in the present conditions driverless cars would affect negatively to the traffic system performance. In other words, congestion would increase proportionally to the number of driverless vehicles in the infrastructure. Consider for example the car following headway between autonomous vehicles, currently being designed around 2s in order to achieve a safe and smooth driving. This leads to a capacity of 1800 veh/h/lane, significantly lower than the typical 2300 veh/h/lane achieved in standard freeways. Equally conservative parameters are proposed for the acceleration, aggressiveness, gap acceptance, lane changing and merging thresholds, etc. Furthermore, several surveys reveal that users are not willing to accept aggressive behavior of self-driving cars, even if it was safe. Besides, decisions taken by individual vehicles on their own (e.g. lane selection, travelling speed, headway, spacing, lane changing, ) would not lead to the global efficiency of the system. In this context, cooperative driving appears as a solution. Autonomous vehicles need to cooperate in order to be globally efficient. Cooperative management of driverless cars would allow creating dense vehicle platoons, with very small spacings emulating a physical linkage between them and creating road trains. These autonomous platoons may coexist with traditional vehicles in the same lane, but still the potential capacity increase is huge, provided that the travelling speed is above some threshold (e.g. 50 km/h). Otherwise, the performance increase in mixed lanes would be greatly reduced. In congested conditions, with stop&go driving and very low average speeds, dedicated lanes might be necessary. They can be shared with other types of privileged vehicles (e.g. HOV, buses, motorbikes, public services, low emissions, etc.) in order to increase the usage of these lanes in the early stages of implementation of driverless cars. Platoon lane selection, traveling speed, composition (cars/trucks), mixed or dedicated strategies, or their effect at junctions are topics to be analyzed in the present project. Traffic modeling is necessary when real life experiments are not feasible. In this project macroscopic modeling is proposed. This allows reproducing the mixed lane behavior for different penetration levels of autonomous vehicles and platooning. The result will be a robust model, based on few parameters, allowing to identify the existing trade-offs and the best management strategies. This analytical model will be programmed into a mesoscopic simulator, based on the Cell Transmission Model per lane and with different types of vehicles. The simulator will ease the assessment of the proposed strategies on realistic environments. The final objective of the COOP project is then to define a set of strategies and guidelines for the cooperative management of driverless cars in a mixed environment with traditional vehicles. With such management, autonomous driving will be, not only safe and convenient, but also efficient.
Plan Estatal de Investigación Científica y Técnica y de Innovación 2013-2016
Programa Estatal de I+D+i Orientada a los Retos de la Sociedad
Retos de Investigación: Proyectos de I+D+i
Gobierno De España. Ministerio De Economía Y Competitividad, Mineco