Vector-based pedestrian navigation in cities

Area of Science:

• Urban planning
• Human mobility
• Computational social science

Background:

• Understanding pedestrian navigation in urban environments is crucial for city planning and transportation.
• Previous research relied on controlled experiments, lacking real-world mobility data insights.

Purpose of the Study:

• To analyze human path planning in city street networks using real-world GPS data.
• To identify key factors influencing pedestrian route selection beyond shortest distance.

Main Methods:

• Statistical analysis of a large dataset of GPS traces from two major US cities.
• Development and testing of a novel vector-based navigation model ('pointiest paths').

Main Results:

• Pedestrians increasingly deviate from the shortest path as origin-destination distance increases.
• Path choices differ significantly when origin and destination are reversed.
• The 'pointiest paths' model statistically outperforms shortest-path models with stochastic effects.

Conclusions:

• Direction to the goal is a primary driver in human path planning.
• Vector-based navigation appears to be a universal property of human path planning across different urban networks.