Multi-Agent Modeling, Simulation & Analysis of Mixed Vehicular Traffic of Autonomous & Human Driven Vehicles and of Spread of Epidemics & their Mitigation Strategies

多智能体建模、仿真

• 批准号: RGPIN-2021-03281
• 负责人: Lawniczak, Anna
• 金额: $ 1.31万
• 依托单位: University of Guelph
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2021
• 资助国家: 加拿大
• 起止时间: 2021-01-01 至 2022-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=742474
• 关键词: Multi; Agent; Modeling; Simulation; Analysis

The objective of the proposed research is the further development & investigation of individually based modeling methodologies and their application to two streams: (1) study of the dynamics of highway traffic consisting of mixed regular (human driven) vehicles (RV) and autonomous vehicles (AV); (2) spread of epidemics of infectious diseases (e.g., COVID-19, COVID-19 & influenza) in realistic population distributions as per Statistics Canada census data and the study of the effectiveness of non-pharmaceutical interventions and vaccination strategies. Thus, the objectives are: (1) the development of large scale Multi-Agent Based Simulation (MABS) models with reactive and cognitive agents; (2) extensive analysis of how their quantitative & qualitative parameters/factors, their levels and their interactions affect the dynamics & performance; (3) application of statistical and machine learning methods for control & prediction of their dynamics & performance; (4) development of novel statistical & machine learning methods as needed by the data analysis of the MABS models. The novelty of the proposed research is in studying the mixed highway traffic consisting of AV & RV, which is predicted to persist at least until 2055. Currently, the focus of research on AV is mostly on individual vehicles. The research on the hybrid vehicular traffic (RV, AV) is in its infancy, but it is important to conduct such research as consequences of such hybrid traffic are unknown and they will affect highway traffic and its management for the next 30-40 years. The impact of this research is in providing the traffic engineers with new knowledge of how various mixes of AV & RV affect traffic flow & congestion, thus roads/highways capacity and in turn transportation infrastructure efficiency. These will provide traffic engineers the information needed for the management of highway traffic. Epidemics of infectious diseases (e.g., COVID-19), for which pharmaceutical interventions do not exist, are a great threat to public health & economy. The development of efficient mitigation strategies is very challenging due to various uncertainties and non-linearities & complexities of epidemics. The novelty and significance of this proposal is in: (1) the development of large scale, MABS models (with reactive agents, cognitive agents) to test the modelling of various non-pharmaceutical interventions & vaccination strategies for COVID-19 (and other epidemics) in large scale realistic population distributions (e.g., Southern Ontario); (2) the development of a deep reinforcement learning approach to automatically learn strategies to control spread of epidemics. In conclusion, the new knowledge produced by the proposed research will impact two communities: traffic engineers, public health practitioners & decision-makers by giving them means to understand these complex systems, their phenomena and to improve their operations.

拟议研究的目标是进一步开发和调查基于个体的建模方法及其在两个流中的应用：（1）研究由混合常规（人类驾驶）车辆（RV）和自动驾驶车辆（AV）组成的公路交通动态;（2）传染病流行病的传播（例如，根据加拿大统计局的人口普查数据以及对非药物干预措施和疫苗接种策略的有效性的研究，在现实的人口分布中，COVID-19，COVID-19和流感）。因此，目标是：（1）开发具有反应和认知智能体的大规模基于多智能体的仿真（MABS）模型;（2）广泛分析它们的定量和定性参数/因素、它们的水平和它们的相互作用如何影响动态和性能;（3）应用统计和机器学习方法来控制和预测它们的动态和性能;（4）根据MABS模型数据分析的需要，开发新的统计和机器学习方法。 拟议研究的新奇在于研究由AV和RV组成的混合公路交通，预计至少持续到2055年。目前，对自动驾驶汽车的研究主要集中在单个车辆上。混合车辆交通（RV、AV）的研究尚处于起步阶段，但其后果尚不清楚，将影响未来30-40年的公路交通及其管理，因此开展此类研究具有重要意义。这项研究的影响是为交通工程师提供了关于AV和RV的各种混合如何影响交通流量和拥堵的新知识，从而影响道路/高速公路的容量，进而影响交通基础设施的效率。这些将为交通工程师提供管理公路交通所需的信息。 传染病的流行（例如，新型冠状病毒（COVID-19）是对公共卫生和经济的巨大威胁，对此不存在药物干预。由于流行病的各种不确定性、非线性和复杂性，制定有效的缓解战略非常具有挑战性。该提案的新奇和重要性在于：（1）开发大规模MABS模型（具有反应剂，认知剂），以测试大规模现实人口分布（例如，南安大略）;（2）开发深度强化学习方法，以自动学习控制流行病传播的策略。总之，拟议研究产生的新知识将影响两个社区：交通工程师，公共卫生从业者和决策者，为他们提供了解这些复杂系统及其现象并改善其运营的方法。