RONIN: A SUMO Interoperable Mesoscopic Urban Traffic Simulator

Road traffic prediction is a major challenge for urban operators and planners alike, as it is difficult to model (drivers' behaviours being complex). Different types of urban traffic simulators have been proposed in the past: macroscopic simulators, fast but not particularly accurate; microscopic simulators, accurate but slow; and mesoscopic simulators, a trade-off between macroscopic and microscopic simulators. The challenge is that a good simulator has to be both accurate and fast, as the prediction tasks required by traffic operators and planners often need to be done quickly. In this paper, we present RONIN, an open-source traffic simulator that is both orders of magnitude faster than the state-of-The-Art microscopic simulator (SUMO) while being more accurate than a typical macroscopic simulator. RONIN is designed as a mesoscopic urban traffic simulator and exploits the location of individual vehicles while also only estimating their trajectories. Besides, unlike most simulators, RONIN is interoperable with a microscopic simulator (i.e., SUMO) by design thus allowing the operators to mix microscopic and mesoscopic speed/accuracy in their traffic management.