Creating a Data Science Competition for Intersection Safety Systems — Insights from the U.S. DOT Intersection Safety Challenge Stage 1B System Assessment and Virtual Testing

Details

• Creators:
  Goli, Mohammad ; Townsend, Haley ; Scarano, Stephen ; Seshadri, Anand ; Young, Caden ; Silverstein, Claire ; Wang, Peiwei ; Wunderlich, Karl
• Corporate Creators:
  Noblis, Inc.
• Corporate Contributors:
  United States. Department of Transportation. Intelligent Transportation Systems Joint Program Office
• Subject/TRT Terms:
  Artificial Intelligence Calibration Data Fusion Evaluation Intersections Machine Learning Pedestrians Predictive Models Safety Sensors Transportation Safety

  More +
• Publication/ Report Number:
  FHWA-JPO-26-005
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States
• Edition:
  Final Report
• Contracting Officer:
  Ocel, Norah
• Corporate Publisher:
  United States. Department of Transportation. Intelligent Transportation Systems Joint Program Office (ITS JPO)
• Abstract:
  To assess the technological maturity of data fusion and artificial intelligence (AI) capabilities of intersection safety systems (ISS), the U.S. DOT designed the Intersection Safety Challenge Stage 1B: System Assessment and Virtual Testing as a data science competition. For this data science competition, the U.S. DOT collected and provided real-world sensor data from a controlled test intersection at the Federal Highway Administration’s (FHWA’s) Turner-Fairbank Highway Research Center (TFHRC) in McLean, VA. The data was collected over several months of operation in 2023 and 2024 from multiple roadway sensors of different types (e.g., camera, Light Detection and Ranging [LiDAR], radar, thermal camera) and covered multiple scenarios, including non-conflicts, potential conflicts, and actual collisions between vehicles and pedestrians. Please note that surrogate pedestrians were used when necessary, and no one was harmed in the data collection process. The purpose of this document is to describe the impetus, assumptions, practical considerations, evolution, and lessons learned for designing and facilitating a data science competition for AI-based ISS, grounded in experiences from the U.S. DOT Intersection Safety Challenge Stage 1B.
• Format:
  PDF
• Funding:
  693JJ321D000021
• Collection(s):
  ITSJPO Digital Library Federal Highway Administration
• Main Document Checksum:
  urn:sha-512:ce2a6e5e6df993e52e8113aa441f5226f869597e0f0a889431513da44610df13b5bffebb8755e2711de534d12884ca384bd767d6b542e5bb25ab5638bc12b575
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/89361/dot_89361_DS1.pdf
• File Type:
  [PDF - 954.55 KB ]