Biblio

The introduction of connected and autonomous vehicles enables new possibilities in vehicle routing: Knowing the origin and destination of each vehicle in the network can allow for coordinated real-time routing of the vehicles to optimize network performance. However, this relies on individual vehicles being “altruistic,” i.e., willing to accept alternative less-preferred routes. We conduct a study to compare different levels of agent altruism in decentralized vehicles coordination and the effect on the network-level traffic performance. This work introduces novel load-balancing scenarios of traffic flow in real-world cities for varied levels of agent altruism. We show evidence that the new decentralized optimization router is more effective with networks of high load.

Driver assist features such as adaptive cruise control (ACC) and highway assistants are becoming increasingly prevalent on commercially available vehicles. These systems are typically designed for safety and rider comfort. However, these systems are often not designed with the quality of the overall traffic flow in mind. For such a system to be beneficial to the traffic flow, it must be string stable and minimize the inter-vehicle spacing to maximize throughput, while still being safe. We propose a methodology to select autonomous driving system parameters that are both safe and string stable using the existing control framework already implemented on commercially available ACC vehicles. Optimal parameter values are selected via model-based optimization for an example highway assistant controller with path planning.