Explaining Deep Learning Decisions to Improve Cognitive Trust of Autonomous Driving

Autonomous driving technology holds significant promise for transforming daily transportation, but societal concerns over its reliability remain high. This research project, led by Dr. Jinbo Bi at the University of Connecticut, aims to enhance cognitive trust in autonomous vehicles (AVs) by developing methods to explain the decisions made by deep learning (DL) models within AV systems. Using explainable AI (XAI) techniques, the project addresses the "black box" nature of DL models, offering transparency in how AVs interpret sensor data, process it, and make real-time driving decisions. This initiative aligns with guidelines, like the EU’s "right of an explanation" and the U.S. AI Bill of Rights, which emphasize user awareness and understanding of AI-driven outcomes impacting consumer safety. The project introduces two innovative XAI approaches: one leverages diffusion models to translate visual inputs into natural language explanations, providing accessible insights into AV actions, while the other identifies critical feature interactions affecting decision-making. These advancements are expected to improve human-centered, accountable AV design, thereby fostering greater public trust and acceptance. By validating these methods through benchmark datasets and real-world simulations, the research also aims to support future AV safety standards and potentially enable technology commercialization, enhancing public and regulatory confidence in AV systems.

Language

• English

Project

• Status: Active
• Funding: $164000
• Contract Numbers:

  69A3552348301

• 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:

  University of Massachusetts, Amherst

  Department of Civil and Environmental Engineering
  130 Natural Resources Road
  Amherst, MA  United States  01003
• Performing Organizations:

  University of Connecticut, Storrs

  Connecticut Transportation Institute
  270 Middle Turnpike, Unit 5202
  Storrs, CT  United States  06269-5202
• Principal Investigators:

  Bi, Jinbo

• Start Date: 20240901
• Expected Completion Date: 20250831
• Actual Completion Date: 0
• USDOT Program: University Transportation Centers Program
• Subprogram: University Transportation Centers

Subject/Index Terms

• TRT Terms: Autonomous vehicles; Safety
• Subject Areas: Data and Information Technology; Highways; Safety and Human Factors;

Filing Info

• Accession Number: 01935902
• Record Type: Research project
• Source Agency: New England University Transportation Center
• Contract Numbers: 69A3552348301
• Files: UTC, RIP
• Created Date: Nov 5 2024 6:20AM