Smart crosswalks

Purpose of the study

The goal of this study is to understand how mediating smart artifacts can promote cooperation and collaboration. Together with other studies described in this website, this one builds into the understanding of artificial mediation of positive social interactions. An extended description is available at artifact mediation studies.


Crosswalks are a natural place for non verbal cooperation or conflict.  In this study we observed and video recorded the behaviors of pedestrians walking in the wild, in groups or by themselves. Next, the activity was re-created in the lab along with the addition of a smart crosswalk which dynamically prompted the ideal distribution of space for the pedestrians on the crosswalk. The study’s final goal was to compare the results of mediated to unmediated in-lab observations and to infer the extent to which the smart crosswalk affected the pedestrians’ interaction.

The study design included two visual conditions with two kinds of crosswalk mediation each, and two controls:

  1. Pedestrians with full perception and smart crosswalks signaling conflicts
  2. Pedestrians with full perception and smart crosswalks suggesting trajectories
  3. Pedestrians with limited perception and smart crosswalks signaling conflicts
  4. Pedestrians with limited perception and smart crosswalks suggesting trajectories
  5. Control with full perception
  6. Control with limited perception

The most frequent configurations of pedestrians on a crosswalk observed in the wild were described using hypergraphs in which nodes represent people and artifacts and silhouettes represent the collectives of human-nonhumans present in a given interaction.

Simplified actor-network hypergraph. Configuration of pedestrians 1 vs. 2. P1, P2 and P3 represent one pedestrian each and CW stands for smart crosswalk

The footages were further analyzed with the help of Path Analytics, a custom made software for the identification of trajectories and the assessment of average stride speed, direction and group cohesion of multiple pedestrians. The software also forecasted potential areas of conflict between opposing walkers in running time.


Path Analytics screen shot depicting pedestrians’ trajectories on a crosswalk

Real-life situations were replicated in the lab with and without the intervention of a smart crosswalk. The smart crosswalk’s function was to identify potential conflict areas on the ground and dynamically prompt the ideal distribution of space for current pedestrians.

Path Analytics was used in the lab to study the behavior of pedestrians in every condition and in several configurations of pedestrian groups, for instance, one pedestrian heading south and two pedestrians heading north as pictured in the image below.

Analysis of interactions in the lab with Path Analytics

Three dependent variables were recorded and analyzed statistically to determine the effects of the mediating crosswalk in the interaction of pedestrians. The figure below presents an example of the analysis of variances per each configuration of pedestrians group.

Sidewalk rounds full signaling-01Results

Statistical analysis of stride speed, direction and trajectory deviation.

Between interactions
Reynolds‘ model of swarm behavior defines separation, cohesion and alignment as the three simple rules governing the locomotive behavior of flocks, schools or swarms of organisms with no central coordination. Although crowds of people tend to follow these rules, our behavior while pursuing individualistic interests is affected by moral principles as observed in Manhattan’s crowd of shoppers. In Tokio, Docomo has studied the risks of dumbwalking while texting on your smartphone illustrating how limited perception of the context increases social viscosity.

This case study is discussed in detail in my doctoral dissertation.