SBIR Phase II: Information fusion-driven adaptive corridor-wide traffic signal re-timing

这个小型企业创新研究(SBIR)第二阶段项目的更广泛影响集中在自适应交通信号配时解决方案上。世界各地的城市和市政当局每年花费超过40亿美元来重新调整交通信号灯的时间，但往往无法充分减少道路拥堵。交通信号灯计时不当的后果是：a)由于拥堵导致的生产率损失增加，美国人平均每年在交通堵塞中停留97个小时；b)交通事故增加，美国道路上每15分钟就有一人死亡；c)温室气体排放增加，三分之一的温室气体排放是由道路上的车辆造成的。该项目将支持基于网络的技术的开发和商业化，以支持城市和市政当局的交通管理人员使用人工智能(AI)和大数据分析更好地管理交通。除了改善交通流量和减少拥堵，该系统还将显著减少有害排放，导致更环保的城市街道。这项技术在美国和欧洲的可用市场总共占全球信号计时市场的60%，这代表着24亿美元的机会。这个小型企业创新研究第二阶段项目寻求开发一种完全自适应的交通信号重新配时解决方案的概念验证，该解决方案可以稳健地处理多个信号走廊以实现商业化。这一努力的关键智力价值将是开发深度学习模型，该模型可以大规模运行，并稳健地处理传感器噪声。强化学习过程将帮助系统适应不同规模的不断变化的交通情景，而不需要人工干预。在第二阶段必须克服的研究目标集中在：1)扩展解决方案；2)使解决方案健壮；以及3)确保系统为用户准备好。实现这些目标可能有助于确保产品能够在经济地部署在云上的大数据架构上成功运行。该解决方案还将提供对人机交互的更深入理解。总体而言，拟议的系统可能会减少信号计时系统的实施时间以及资本和维护费用。这些优势将鼓励美国和国际上的城市采用这种信号计时策略。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact of this Small Business Innovation Research (SBIR) Phase II project focuses on an adaptive traffic signal timing solution. Cities and municipalities worldwide spend over $4 billion annually to retime traffic signals and yet often fail to adequately reduce congestion on roadways. The consequences of mistimed traffic signal timing are: a) increasing productivity losses due to congestion with the average American spending 97 hours stuck in traffic every year, b) increasing accidents due to traffic, with one fatality every 15 minutes on US roads, and c) increasing greenhouse emissions with a third of all emissions caused by vehicles on the roads. This project will support the development and commercialization of a web-based technology to support traffic managers in cities and municipalities to better manage traffic using artificial intelligence (AI) and big data analytics. In addition to improving traffic flow and reducing congestion, the system will also significantly reduce harmful emissions, leading to more environmentally friendly city streets. The serviceable markets for this technology in the US and Europe, which together constitute 60% of the global signal-timing market, represent a $2.4 billion opportunity. This Small Business Innovation Research Phase II project seeks to develop a proof-of-concept for a fully adaptive traffic signal retiming solution that can robustly handle multiple signal corridors for commercialization. The key intellectual merit of this effort will be developing deep learning models that can run at scale and handle sensor noise robustly. The reinforcement learning process will help the system to adapt to changing traffic scenarios at different scales without the need for manual interventions. Research objectives that must be overcome in Phase II are focused on: 1) scaling the solution; 2) making the solution robust; and 3) ensuring that the system is user ready. Achieving these objectives may help ensure the product can successfully run on big-data architecture economically deployed on the cloud. The solution will also provide a deeper understanding of human-machine interaction. Overall, the proposed system may reduce implementation time as well as capital and maintenance expenditures for signal timing systems. These advantages will encourage cities around the US and internationally to adopt such signal timing strategies.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.