Cruising for parking is a problem that cities have long struggled to address. Recent efforts by the FHWA in San Francisco, Minneapolis, Los Angeles, and other cities have advanced the state of knowledge but large gaps remain. This project seeks to address these important gaps by using GPS traces, GPS-augmented travel surveys, and GPS traces with driver video to quantify the extent of cruising, and understand the characteristics of cruising behavior. We will develop two sets of algorithms to better explain the physical and psychosocial dimensions of the phenomenon: (1) Driver behavior discerned from geographic patterns of driving (vehicle tracking: speed changes and circling), and (2) Driver behavior discerned from changes in outward appearance (person tracking: eye and head movements, combined with vehicle tracking). We will cross reference and calibrate the algorithms with historical sensor data and experimental GPS traces developed in the context of this project. The objective is to identify data strategies to exploit the increasingly available, yet disparate, large data sources that can be brought to bear. Future solutions will inevitably involve integrating these sources across a single data platform, and we will provide a prototype data fusion platform.
