Explainable AI Algorithm for Driving Decision-Making

Explainable AI Algorithm for Driving Decision-Making

Last Updated: 01/22/2025 | All information is accurate and up-to-date

Key Features

  1. Implicit visual-semantic module captures the region-based action-inducing components (implicit visual semantics) to provide a human-understandable explanation;
  2. Explicit reasoning module is developed to jointly align the human-annotated explanation and intention prediction in a multi-task fashion.
A diagram explaining how the AI algorithm makes driving decisions.

Multi-modal Feedback for Explainable Driving Decision-making

A diagram showing multi-modal feedback for explainable driving decision-making

Key Features

  1. Multi-head-attention-based self-learning scene graph;
  2. Adjustable balance between global and local features, which determines automatic and human-guided feature weighting.
A multi-head-attention-based self-learning scene graph
A chart that shows the increased effects of the Interrelation module.
  • Action prediction accuracy is optimal at balanced λ;
  • Explanation accuracy always increases.
Several diagrams explaining the researchers use of multi-task learning boosting.
Multi-task learning boosting