The curb space is the portion of the public rights-of-way that demarcates the roadway from the sidewalk, separating pedestrian flow from moving vehicles. It is a scarce public resource that has been traditionally used for storing private passenger vehicles. However, the past decade has seen not only a surge in demand but also the rise of new demands for curb space, driven by new forces of change: the rise in online shopping has driven up the demand for delivery vehicle loading and unloading spaces; the increasing use of ride-hailing vehicles such as Uber and Lyft has exacerbated curb space congestion; the rapid adoption of micromobility modes has increased their parking demand, among others. The pandemic has only exacerbated the issue due to greater demand for home delivery services and novel use cases such as curbside cafes.
The mismatch between the increase in demand and the lack of curb space supply represents a bottleneck in the urban transportation system, increasing the cruising for parking time — the time drivers spend searching for parking — as well as the occurrence of unauthorized parking. Both consequences heavily impact urban traffic congestion, increasing emissions and lowering the quality of life for urban dwellers, as well as can potentially create unsafe conditions. More broadly, the curb is a major linchpin in city operations: beyond congestion, it also affects business district vitality, residential access, and even policy decisions about new constructions.
To address these challenges, cities need greater access to data science and machine learning tools to have better insights into the overall use of and demand for curb space, with the final objective to be able to effectively manage the limited amount of curb space available. This includes the need for tools to aid in optimizing pricing mechanisms and to adaptively learn the most efficient and sustainable allocation of space to the different types of users.
Two research groups at the University of Washington have taken different but complementary approaches to study the curb and build tools to help cities understand different curb demands and better manage the limited curb space available.
The Urban Freight Lab, led by Prof. Anne Goodchild, approaches the study of the curb from the perspective of commercial vehicles, including delivery and ridehailing vehicles. The group has collected data and derived statistical models of curb users’ behaviors for commercial vehicles. Furthermore, the group has piloted on-the-ground technologies and policies to improve curb access. In a recent project, Prof. Goodchild’s group deployed 300 in-ground occupancy sensors at commercial vehicle load zones (CVLZs) and passenger load zones (PLZs) — curb spaces dedicated to commercial and ridehailing vehicles — in a 10-block study area in the Belltown neighborhood of Seattle, WA, collecting more than a year of fine-grained curb-use data.
The research group led by Prof. Lilian Ratliff approaches the study of the curb primarily from the perspective of private passenger vehicles, applying innovative machine learning and game theory tools to study curb management policies. In a recent project, Prof. Ratliff’s group developed a new modeling framework to estimate on-street paid parking occupancies — spaces dedicated for private passenger vehicle parking — from parking transaction data and sparse ground truth occupancy data obtained via manual counts and timelapse camera images.
The research in Goodchild’s and Ratliff’s groups has been impactful. Yet, load zone and paid parking curb-uses are highly interdependent given that the zones dedicated to the different use cases are often on the same curb. Hence, a more holistic approach to learning curb use behaviors is needed in order to effectively manage the whole curb.
For this project, the two groups will collaborate to integrate different data streams currently being collected separately and in an uncoordinated way, including data from in-ground curb sensors at CVLZs and PLZs, paid parking transactions at paid parking spaces, and data obtained from timelapse camera recordings. With such a complete dataset, the groups will create a holistic framework to analyze not only the curb behaviors of different users but also how different users interact in the competition for limited curb space.
The proposed collaboration will advance the state of the art in environmental sciences by providing the most complete dataset and creating innovative tools to inform policymaking on curb parking pricing and curb allocation to reduce cruising for parking and unauthorized parking events, therefore tackling the climate crisis by reducing urban vehicle emissions and traffic congestion.
The proposed collaboration will also advance the state of the art in data science by developing a new statistical framework and machine learning algorithms to analyze curb space use behaviors from different curb space users and develop much-needed recommendations for cities on how to better allocate curb space to different competing demands.
The project will have a direct impact on the City of Seattle as both groups are currently collaborating with the Seattle Department of Transportation to create a more data-driven decision-making framework for curb space policies, as well as an impact on the fields of urban transportation and logistics by merging two separate kinds of literature, the more traditional transport theory taking private passenger vehicles as the main actor in urban transportation and the urban logistics field that focuses on commercial vehicles operations in urban areas.
Concrete outcomes of the projects obtained during the year of collaboration will include a joint seminar series of the two groups, presenting their ongoing research projects that focused on the curb, a join effort to collect data in Seattle, and integrating data streams to generate a complete dataset of curb use for the Seattle downtown area. Additionally, the groups will jointly write a scientific paper proposing a holistic framework to analyze the curb from the different users’ perspectives. The proposed collaboration will expand upon the projects Prof. Goodchild’s and Prof. Ratliff’s groups are currently working on, and develop a new set of data and tools that will enable future joint grant proposals by the two groups.