The Future of HD Mapping: Crowdsourcing to Improve PNT Resilience and Safety

Positioning, Navigation, and Timing (PNT) systems are essential in Highly Automated Transportation Systems (HATS), and HD Maps provide critical information to both Advanced Driver Assistance Systems (ADAS) and Autonomous Driving technologies. These maps go beyond basic navigation, offering essential spatial information, such as precise details, including road edge lines, lanes, freeway exits, emergency and HOV lanes, crosswalks, and various infrastructure features like traffic signs, bridges, tunnels, etc., typically with high accuracy. The adoption of HD Maps is increasing as HATS deployments grow, and HD Maps are also expanding to include high-level information such as lane-specific rules, traffic light schedules, and more. Car manufacturers currently create their HD Maps, leading to duplication of efforts and a lack of standardization. Although initiatives like OpenDRIVE aim for industry standards, proprietary approaches are still prevalent. While regulating the HD Map format in the car manufacturing industry may be late, there is potential for consolidation in the future. In addition, HD Maps have not yet been fully exploited, and there is room for improvement in terms of handling PNT vulnerabilities and enhancing safety. Objectives: The primary goal is to create a concept for the future of HD Maps, including their creation, updating, and dissemination. The vision includes live HD Maps based on crowdsourcing as the primary source of geospatial data. Government authorities are expected to manage live HD Maps as a service, allowing real-time data exchange with vehicles. In the research team's view, HD Maps hosted in ITS centers should deliver all the information to the vehicles on-the-fly as they move in the transportation network as well as receive new sensor data coming from vehicles that are aggregated and used for HD Map updates. The main challenge lies in continuously updating the HD Map system based on data collected by HATS vehicles. Detecting changes from moderately accurate but highly redundant data requires AI technologies due to the complexity, size and non-linearity of the data. Data from HATS vehicles can provide valuable information for HD Maps, but transmitting all sensor data is impractical. An important research question is to find the balance between what should be processed at the vehicle level and what should be sent to a central processing center. AI/ML/DL technologies offer solutions to tackle both problems. Contrastive learning is proposed as a method for change detection in HD Maps. Contrastive learning is a type of unsupervised learning that aims to learn representations of data by maximizing the similarity between positive pairs (similar items) and minimizing the similarity between negative pairs (dissimilar items). In this context, it may be used to detect changes or differences between the existing HD Map and the real-time data collected from the environment. Reinforcement learning is suggested for incremental updates to the HD Map. Reinforcement learning involves training an agent to make sequential decisions in an environment to maximize a reward signal. In this case, it could be used to determine how the HD Map should be updated based on changes detected through contrastive learning or other means.


    • English


    • Status: Active
    • Funding: $112926
    • Contract Numbers:


    • Sponsor Organizations:

      Office of the Assistant Secretary for Research and Technology

      University Transportation Centers Program
      Department of Transportation
      Washington, DC  United States  20590
    • Managing Organizations:

      Center for Automated Vehicle Research with Multimodal Assured Navigation

      Ohio State University
      Columbus, OH  United States  43210
    • Project Managers:

      Kline, Robin

    • Performing Organizations:

      Ohio State University Center for Automotive Research

      930 Kinnear Road
      Columbus, OH  United States  43212
    • Principal Investigators:

      Toth, Charles

    • Start Date: 20231030
    • Expected Completion Date: 20240830
    • Actual Completion Date: 0
    • USDOT Program: University Transportation Centers Program

    Subject/Index Terms

    Filing Info

    • Accession Number: 01901381
    • Record Type: Research project
    • Source Agency: Center for Automated Vehicle Research with Multimodal Assured Navigation
    • Contract Numbers: 69A3552348327
    • Files: UTC, RIP
    • Created Date: Dec 4 2023 6:50PM