Approximately orchestrated routing and transportation analyzer: City-scale traffic simulation and control schemes
MetadataShow full item record
Along with other intelligent traffi c control schemes, autonomous vehicles present new opportunities for addressing traffi c congestion. Traffi c simulators enable researchers to explore these possibilities be- fore such vehicles are widespread. This thesis describes a new open source, agent-based simulator: the Approximately Orchestrated Routing and Transportation Analyzer, or AORTA. AORTA is designed to provide reasonably realistic simulation of any city in the world with zero con figuration, to run on cheap machines, and with an emphasis on easy use and simple code. Experiments described in this thesis can be set up on a new city in about five minutes. Two applications are built on AORTA by creating new intersection control policies and specifying new strategies for routing drivers. The first application, intersection auctions, allows humans to instruct their autonomous vehicle to bid on their behalf at intersections, granting them entry before other drivers who desire conflicting movements. The second, externality pricing, learns the travel time of a variety of di fferent routes and defi nes a localized notion of cost imposed on other drivers by following the route. This information is used to tax drivers who choose to improve their own trip by inconveniencing others. These two systems demonstrate AORTA's utility for simulating control of the traffi c of the near future.