V2V-Based Memetic Optimization for Improving Traffic Efficiency on Multi-Lane Roads

Within the next few years, autonomous vehicles will start being commercialized. In the same time, inter-vehicular communication is emerging. Using these new technologies allows designing new software to make cooperative cars. These cooperatives cars can exchange messages to improve traffic efficiency by, for instance, notifying about the presences of other cars, or managing the right-of-way at intersections. In the context of multi-lane roads, we propose to use the communication between vehicles to design a cooperative intelligence, based on evolutionary optimization where the behavior of each vehicle is regularly updated according to the behavior of surrounding vehicles and a fitness function. This paper presents an automated/cooperative lane-change framework where the parameters of the system are dynamically adjusted using an online evolutionary algorithm. The goal is to make the cars adjust their behavior according to the local traffic conditions. Simulations are carried out showing a performance improvement in terms of traffic fluidity.