Explainable Edge Computing in a Distributed AI - Powered Autonomous Vehicular Networks

In the ever-dynamic landscape of autonomy and un-paralleled intelligence, the vehicles moving across any geospatial terrain create an Artificial Intelligence (AI)-powered vehicular network that can interact quickly to make autonomous decisions. However, the efficient working of autonomous vehicles (AV s) relies on seamless data integration from various resources, including Vehicle-to-vehicle (V2V), Vehicle-to-Infrastructure (V2I), and other communications in the AI-powered networks. In such scenarios, the velocity and pattern of vehicle mobility and motion require high effectiveness and efficiency in AI-enabled decision-making. Additionally, understanding the critical basis or feature data that triggers a particular decision from an AI model is also essential to enhance user acceptance and trust. This paper introduces an explainable edge computing-based solution to enhance the performance of AVs in an AI-powered vehicular network. The solution focuses on harnessing the power of local edge through Explainable Artificial Intelligence (XAI). This data is sourced from local edge computing nodes and is efficiently disseminated to the global edge. A comprehensive approach is designed for identifying and amalgamating important feature data from the local edge to the global edge. Through this cooperative approach, autonomous vehicular networks attain elevated efficiency, accuracy, and adaptability, making the driving experience of an AV more reliable and secure.