Accurate assessment and prediction of pedestrian flows over the urban street network could enable data-driven transportation and city planning for promoting walking and other planning goals related to pedestrian flows. Since the available data collection methods are limited, and too costly to implement on a large urban scale, it is practically unfeasible to measure pedestrian flows across an entire urban area. The alternative is to use pedestrian flow models. We have developed a pedestrian flow model at large-scale urban environments using Agent-Based Modeling (ABM) approach to assist in predicting pedestrian flows in an existing and planned urban street network.

UNMET NEED
Current agent-based models of pedestrian flow (of key players available in the market) were mostly applied for modeling of crowd flows and pedestrian evacuations in indoor environments and small-scale urban areas, such as: airports, railway stations, malls, stadiums or public square. These models are not able to offer high-resolution simulation and prediction of pedestrian mobility at large scales such as a neighborhood, city region, entire city.

OUR SOLUTION
Our ABM offers a high-resolution simulation and prediction of pedestrian mobility at large scales – neighborhood, city region, entire city. To develop our ABM: (1) We use high-resolution spatio-temporal data at the urban scale; (2) We apply advanced Machine Learning methods to translate empirical data (based on travel behavior survey and GPS-based walking routes) into the general rules of pedestrian behavior and decision-making; (3) We calibrate and validate the ABM in variety of urban areas differ in their morphological, functional and socio-demographic characteristics. Because of this, we have unique knowledge about the walking behavior attributes relevant to pedestrian flow in variety of large-scale urban environments.

APPLICATIONS
Our ABM aims to assist urban and transportation planning in various applications:
• In existing urban environments – pedestrian flows in the urban network for: (1) Detecting discontinuity of pedestrian traffic; (2) Locating roads that require improvement in the infrastructure for pedestrians (e.g., sidewalks and tree planting); (3) Determining the location of land uses (e.g., commercial or park); (4) Assessing the risk level of pedestrians (of different population groups) for road accidents in the urban street network; etc…
• In planned urban environments – evaluating the impacts of actual or planned changes in the urban layout on pedestrian flows in the urban network; examples of changes in the urban layout: (1) Adding walking paths; (2) Establishment of light rail /metro stations; (3) Change/addition of land uses (such as mall or school); (4) Change in housing density; (5) Evaluating the walking activity of residents in a given area; etc…

STATUS
Initial proof of concept – successful validation of the ABM in predicting pedestrian flows in the city Tel – Aviv. Ongoing field tests continue to validate the accuracy and reliability of our ABM in additional cities.

INTELLECTUAL PROPERTY
A provisional patent application is being written