Exploration of Explainable AI in Context of Human-Machine Interface for the Assistive Driving System

This paper presents the application and issues related to explainable AI in context of a driving assistive system. One of the key functions of the assistive system is to signal potential risks or hazards to the driver in order to allow for prompt actions and timely attention to possible problems occurring on the road. The decision making of an AI component needs to be explainable in order to minimise the time it takes for a driver to decide on whether any action is necessary to avoid the risk of collision or crash. In the explored cases, the autonomous system does not act as a “replacement” for the human driver, instead, its role is to assist the driver to respond to challenging driving situations, possibly difficult manoeuvres or complex road scenarios. The proposed solution validates the XAI approach for the design of a safety and security system that is able to identify and highlight potential risk in autonomous vehicles.