ViCARUS: Vision-Centric Connected Autonomy for Vulnerable Road Users' Safety

Details

• Creators:
  Bhagavatula, Vijayakumar ; Shenkut, Dereje
• Corporate Creators:
  Carnegie Mellon University
• Corporate Contributors:
  Carnegie Mellon University. Traffic21 Institute. Safety21 University Transportation Center (UTC) ; United States. Department of Transportation. University Transportation Centers (UTC) Program ; United States. Department of Transportation. Office of the Assistant Secretary for Research and Technology ; United States. Department of Transportation. Federal Highway Administration
• Subject/TRT Terms:
  • Vulnerable road users
  • Highway safety
  • Vehicle to everything communications
  • Object detection
  • Vector analysis
  • Trajectory
  • Mathematical prediction
  • Connected vehicles
  • Autonomous vehicles
  • Algorithms
  More +
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States
• Edition:
  Final Report (July 1, 2025 - June 30, 2026)
• Corporate Publisher:
  Carnegie Mellon University. Traffic21 Institute. Safety21 University Transportation Center (UTC)
• Abstract:
  The safety of vulnerable road users (VRUs) such as pedestrians, cyclists and motorcyclists remains a pressing challenge on U.S. roads. Cooperative perception and prediction, in which connected vehicles and infrastructure exchange sensing information over Vehicle-to-Everything (V2X) communication, can see past the occlusions and range limits of any single vehicle. These capabilities disproportionately benefit VRUs, who are small, easily occluded, and severely injured in collisions, yet communication bandwidth remains the central challenge to deployment. This report presents the results of three studies conducted under the ViCARUS project, covering a collaborative perception method that prioritizes hard, safety-critical instances during communication, a learned feature codec that compresses exchanged features by more than two orders of magnitude with minimal accuracy loss, and a goal-level communication scheme for cooperative trajectory prediction that matches full-trajectory exchange at one tenth of the bandwidth. Together, the results indicate that cooperative perception and prediction for VRU safety is achievable within practical V2X bandwidth budgets.
• Format:
  PDF
• Funding:
  69A3552344811 ; 69A3552348316
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/93523/dot_93523_DS1.pdf
• File Type:
  [PDF - 7.03 MB]
• Collection(s):
  Autonomous Vehicles ; University Transportation Centers
• Main Document Checksum:
  urn:sha-512:578bb66e23f53f32f1a78518d3232e4c6002b2aa1ad297f9b90fe4848a082e32924658bec8ea15dece136a95e42b9f61895b4836d7a5bee6b65fbca9249580c7