SmartSim: An open-source mesoscopic urban traffic simulator for multimodal and reinforcement learning applications

Abstract

City-scale multimodal transport studies require simulation environments that can represent heterogeneous service operations while allowing external algorithms to interact with evolving system states. However, these capabilities remain distributed across existing simulation platforms. To address this gap, this paper presents SmartSim, an open-source, city-scale mesoscopic traffic simulator that operates private vehicles, taxis/ride-hailing vehicles, buses, and rail transit under a unified simulation framework. By representing both fixed-route and demand-responsive services, SmartSim enables heterogeneous transport operations to be simulated concurrently within the same environment. It further adopts a two-layer architecture and a configurable synchronous interface, through which users can define decision points, observations, actions, and execution nodes for external learning and optimization algorithms. Using fused Hong Kong transport data, we construct a full multimodal scenario to evaluate SmartSim. Under the tested demand scales and hardware configuration, SmartSim completed the simulations in less time than MATSim, with lower CPU utilization and memory use. A reinforcement learning based taxi-repositioning experiment further demonstrates the complete closed-loop process of state acquisition, action execution, environment feedback, policy updating, and frozen-policy evaluation over 400 training episodes. An emission-oriented application also illustrates how link-level simulation outputs can support downstream environmental analysis. SmartSim thus provides a common experimental framework for comparing multimodal operational strategies, with potential applications in transport service planning, control evaluation, and road traffic emission assessment.

Keywords

mesoscopic traffic simulator / multimodal traffic simulator / reinforcement learning

Rights & permissions

The Author (s) 2026

Just Accepted

This article has successfully passed peer review and final editorial review, and will soon enter typesetting, proofreading and other publishing processes. The currently displayed version is the accepted final manuscript. The officially published version will be updated with format, DOI and citation information upon launch. We recommend that you pay attention to subsequent journal notifications and preferentially cite the officially published version. Thank you for your support and cooperation.