U.S. DOT Advances Complete Streets Through Artificial Intelligence: A Message from Brian Cronin

The Intelligent Transportation Systems Joint Program Office (ITS JPO) pursues a variety of approaches to foster innovation in the transportation technology research ecosystem; such as challenges, pilots, and small business research investments. In October 2024, the U.S. Department of Transportation (U.S. DOT) awarded $2.4 million in contracts to twelve (12) small businesses as part of the Complete Streets Artificial Intelligence Initiative, a multi-phase effort that seeks to harness Artificial Intelligence (AI) to improve transportation. The goal is to develop powerful new decision-support tools for State, local, and Tribal transportation agencies that assist in the siting, design, and deployment of Complete Streets— streets that prioritize safety, comfort, and connectivity for all road users.

Currently, Federal Aid recipients, other public agencies, and the consultant firms that support them have identified a need for more detailed information about how their transportation networks serve people walking, bicycling, or rolling via micromobility, wheelchairs, or other devices. The Complete Streets AI initiative was created to close these gaps in basic roadway data, such as pedestrian and bicyclist volumes, the presence and condition of sidewalks or bike facilities, and the accessibility of routes for people using wheelchairs or other mobility assistance devices. Harnessing the power of AI to improve data availability and fidelity will allow communities to better plan, design, and implement Complete Streets projects across the country.

This Initiative encourages small businesses to think outside the box. The twelve awardees combine expertise in both AI and transportation. They will utilize a wide range of tools and technologies such as machine learning, computer vision, convolutional neural networks, and generative AI to transform raw data like satellite, aerial, or street-level video and still photography and 3D point cloud information into refined data layers for use in Geographic Information Systems (GIS).

Phase I funding, which launched in October, allows each awardee to create a proof-of-concept at a project pilot location. Phase II will further test and harness these methods in novel data generation and processing methods and scale them to multiple geographic areas. Phase II and III will support the implementation of Complete Streets analysis capabilities in a functioning software solution. Eventually, the software and data generation techniques that the awardees develop may be deployed on a national scale.

The companies and their projects are as follows:

• Aiwaysion, Inc. (Seattle, WA): Artificial Intelligence Approach to Generate and Analyze Complete Streets Data at Scale – Project Location: Tucson, AZ
  This project will create a data collection framework and develop AI and computer vision algorithms to generate and analyze diverse data sets. This framework will identify and address data deficiencies, particularly in areas such as traffic volume, pedestrian infrastructure, and demographic representation, to support more equitable and effective urban planning.

• Creare LLC (Hanover, NH): Data Processing and Analysis Framework for Infrastructure Planning to Promote Active Transport – Project Location: Denver, CO
  This project aims to develop a system that automatically generates user-specified infrastructure data and enables customized, community-specific analyses needed to implement Complete Streets policies. It will use data formats that closely follow existing and emerging standards, providing interoperability with existing GIS analysis tools and pipelines.

• JC-TECHS Corp. (University Place, WA): Scalable and Sustainable AI Solutions for Complete Streets Data Creation and Maintenance with Off-the-shelf Dash Cams – Project Location: Tacoma, WA and Lakewood, WA
  This project will study AI techniques for creating a Complete Street data set utilizing the street videos collected by off-the-shelf dash cams. Objectives include developing routing algorithms and a mobile app for data maintenance, designing secure data storage schemes, and devising ways to demonstrate the accuracy of the collected data.

• Kittelson and Associates (Wilmington, NC): Safe Routes for All – Using AI-Based Image Recognition and Machine Learning Algorithms for Network-Wide Assessment and Routing of Multimodal Trips Based on Level of Traffic Stress – Project Location: Tampa, FL
  This project will use AI-based image processing and machine-learning technologies to fill critical gaps in the data obtained from aerial imagery. The goal is to enhance agencies’ ability to assess their multimodal networks and prioritize infrastructure investments that create safe, low-stress routes for pedestrians and bicyclists.

• Kitware, Inc. (Clifton Park, NY): Complete Urban to Rural Balanced Streets by Artificial Intelligent Design – Project Location: Greater Albany, NY
  The purpose of this project is to automate the generation of walkway networks on a national scale. The first phase of the project will use overhead imagery and computer-vision algorithms to build a detailed, accurate traffic model. The second phase will develop an AI assistant that can respond to queries and help planners visualize network improvements.

• Numobility LLC (Atlanta, GA): Safe Accessible Multimodal Mobility Intelligence for All Road Users – Project Location: Atlanta, GA, Gwinnett County, GA, Bellevue, WA, King County, WA, Austin, TX, and Phoenix, AZ
  This project aims to develop an AI-driven decision support system for identifying and prioritizing Complete Streets projects. The system will integrate diverse data sources on travel behavior, infrastructure conditions, and environmental context to provide actionable insights.

• OpalAI Inc. (Beverly Hills, CA): AI for City – Location: Los Angeles, CA
  This project will develop a scalable, AI-based tool for improving roadway design. Comprehensive data collection, automated analysis, and multi-objective algorithms will help planners make data-driven decisions to support improvement projects that meet Complete Streets criteria.

• Skylite Group (Bethesda, MD): CSAI Phase 1 – Project Location: Washington, DC and Montgomery County, MD
  This project will combine data from both overhead and surface-based imagery with geospatial data to form a more complete picture of the situation on the ground. This high-quality dataset would enable users to rapidly identify unsafe road crossings, analyze the utilization of existing bike lanes, and determine where new bike lanes might be needed.

• State of Place (Natick, MA): AI for Complete, Equitable Streets: Using Computer Vision and Machine Learning to Deliver Data-Driven Guidance for Complete Streets – Project Location: Philadelphia, PA
  This project will use visual machine-learning models to generate street-level data on infrastructure and context features as well as vehicle, bicyclist, and pedestrian traffic. A machine-learning model will use this and other relevant data to predict jurisdiction-wide, multimodal travel patterns to inform the siting, prioritization, design, implementation, and evaluation of proposed improvements.

• TrAnalytics LLC (Bedford, MA): Complete Pavement Markings for Safe and Complete Streets – Project Location: Greater Boston, MA
  This project will demonstrate the feasibility and cost-effectiveness of using large-scale lidar data to automatically assess pavement marking conditions, with a particular focus on safety.

• VELO.AI, Inc. (Pittsburgh, PA): Informing Infrastructure Interventions via Novel Near-Miss Data Collection – Project Location: Pittsburgh, PA
  This project will use a fleet of bicycles equipped with sensors to collect data on “near miss” incidents between bicycles and motor vehicles. This data is expected to enable the development of scalable, predictive models to help infrastructure managers determine where improvements are needed.

• WCEC Engineers, Inc. (Salt Lake City, UT): Complete Street Data Collection and Assessment using Machine Vision – Project Location: Greater Salt Lake City, UT
  The purpose of this project is to develop a robust machine-vision model to identify streetscape assets and define their impact on the Complete Street environment.

To learn more about the Complete Streets AI Initiative, I encourage you to visit the website at its.dot.gov/CSAI. To learn more about the twelve Complete Streets AI Initiative awardees, you can read their full project abstracts here.

As AI technology continues to evolve, the U.S. DOT is working with industry leaders to ensure that AI can be safely implemented to improve our nation’s transportation networks. Together, we can advance safer and more reliable transportation for all road users.

Brian Cronin, Director, ITS JPO

Posted 12/4/24