Vehicle Edge Computing for Travel Behavior and Demand in Future Intelligent Transportation Systems

Details

• Creators:
  Trinidad, Rafael ; Ly, Michael ; Cheng, Yichi ; Wang, Yunsheng ; Zhang, Yongping ; Cheng, Wen
• Corporate Creators:
  California State Polytechnic University, Pomona
• Corporate Contributors:
  United States. Department of Transportation. University Transportation Centers (UTC) Program ; United States. Department of Transportation. Office of the Assistant Secretary for Research and Technology ; National Center for Understanding Future Travel Behavior and Demand (TBD) ; United States. Department of Transportation. Federal Highway Administration
• Subject/TRT Terms:
  Advanced Traffic Management Systems Autonomous Vehicles Cloud Computing Data Analysis Edge Computing Intelligent Transportation Systems Latency Reduction Performance Real-time Processing Real Time Control Travel Behavior Travel Demand

  More +
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States
• Edition:
  Final Report: 2024-2025
• Corporate Publisher:
  United States. Department of Transportation. University Transportation Centers (UTC) Program
• Abstract:
  The future of transportation faces critical challenges due to computational constraints in autonomous vehicles and limitations of cloud-based solutions. This project introduces a vehicular edge computing platform that addresses these challenges by leveraging small-scale autonomous vehicles (Donkey Cars) and edge computing nodes. We implement a dual-protocol communication system where image data streams via UDP while control commands are transmitted through TCP, ensuring both speed and reliability. Our platform demonstrates significant performance improvements in real-time processing and decision-making capabilities, achieving up to 99.75% reduction in inference latency compared to local processing, while reducing CPU usage by 66.5% and memory utilization by 68.5%.A second two-vehicle multi-threaded experiment also demonstrates great performance with CPU usage remaining around 20%, consistent memory usage at 0.8%, and inference latency around 10 milliseconds. The results of both experiments suggest that edge computing integration could be a viable solution for future intelligent transportation systems, particularly in scenarios requiring real-time hazard detection and traffic management. Our results reveal significant reductions in inference latency, with future suggested research focusing on GPU acceleration, CHI@EDGE, and more.
• Format:
  PDF
• Funding:
  69A3552344815 ; 69A3552348320
• Collection(s):
  University Transportation Centers
• Main Document Checksum:
  urn:sha-512:3ff54b23a41c6b3d11d6e73fa956e5dc8376689fb8b4a57272b238daf0d8137ecae2d0b9e50d3014f11f51760b68d0902a0f52dd8cb76f1f52b643876f6a88db
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/91979/dot_91979_DS1.pdf
• File Type:
  [PDF - 461.90 KB ]