Enhancing smart city mobility through real time explainable AI in autonomous vehicles.

Area of Science:

• Artificial Intelligence
• Computer Vision
• Urban Transportation Systems

Background:

• Autonomous Vehicular Networks (AVNs) offer transformative potential for smart cities but face challenges in decision transparency, public trust, and safety.
• Existing AVN development often prioritizes technical reliability over interpretable decision-making processes, hindering public confidence and adoption.
• Lack of understanding in how Autonomous Vehicles (AVs) make real-time decisions impedes broader integration into urban environments.

Purpose of the Study:

• To develop a transparent and interpretable decision-making framework for AVNs using Explainable AI (XAI).
• To enhance the safety, reliability, and public acceptance of AVNs within smart city ecosystems.
• To integrate advanced object detection with AI interpretability for real-time urban mobility.

Main Methods:

• Integration of the You Only Look Once, V5 (YOLOv5) object detection model with Explainable AI (XAI) techniques.
• Development of an XAI-based YOLOv5 model for real-time, explainable decision-making in AVs.
• Evaluation of the model's performance in enhancing transparency, safety, and public confidence.

Main Results:

• The proposed XAI-based YOLOv5 model achieved 99% accuracy with a 1% miss rate.
• Demonstrated enhanced classification accuracy and significant improvements in decision transparency.
• The model effectively addresses the need for interpretable AI in AVN operations.

Conclusions:

• The XAI-based YOLOv5 model provides a robust solution for transparent and explainable decision-making in AVNs.
• Increased transparency and interpretability are crucial for fostering public trust and accelerating AVN adoption in smart cities.
• This research contributes to safer, more reliable, and publicly accepted autonomous transportation systems.