Designing urban areas using traffic simulations, artificial intelligence and acquiring feedback from stakeholders

Cities are very complex systems, people living in urban areas and components of infrastructure are in constant interaction: people travel to work and homes, they use vehicles, roads and other infrastructure components. New, innovative technological ideas frequently emerge, so it is not easy to predict how cities may evolve, nevertheless it is important to plan and build urban infrastructure for a long time horizon, e.g., many years. Even in short-term it is difficult to predict evolution of life in a city (and traffic, in particular), even small initial changes (e.g., car accident) may cause dramatic changes after some time (e.g., formation of traffic jams). Therefore, there is a need for innovative tools, which can help in designing better urban infrastructure (e.g., locations and capacities of parkings or charging stations for electric vehicles) and better algorithms for its usage (e.g., algorithms of drive for autonomous and connected transport or traffic management algorithms). The author proposes a general framework which may help in solving planning and managing problems that urban and traffic engineers have to face or will have to address in the near future, in the era of connected, autonomous, electric and shared transport. The method takes advantage of microscopic simulations and new technologies, such as artificial intelligence, cloud computing / cluster computing, but also feedback from stakeholders, such as municipal authorities, companies (e.g., car manufacturers), professional traffic engineers and citizens. The main idea is that having an accurate microscopic simulation model of vehicular traffic (and, potentially, also models of other complex urban processes), one can evaluate the impact of different (even small) changes on the traffic and life in a city and select settings (e.g., traffic signal settings, locations and capacities of parkings or charging stations, algorithms of drive) which can be optimal according to some objective function (computed using the simulation). However, microscopic traffic simulations are time-demanding, while the space of possible changes (such as changing traffic signal settings, building new road or building new parkings) is very large, so usually it will be computationally intractable to evaluate all possible settings (their number may be astronomical) in order to find the best one. However, the author proposes a few approaches, which combined together may help in solving efficiently such computationally-demanding optimization problems.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01759281
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Nov 12 2020 3:20PM