Fuzzy Cognitive Maps for Intelligent Agent's Artificial Situational Awareness in Collaborative Driving Context

Automation such as the autopilot car technology is considered to be a promising solution to reduce the number of accidents on the road. However, it is believed that to increase safety for such technologies, it is necessary for an intelligent agent (IA) and the human driver to share their situational awareness so that the same conclusions regarding driving situations can be reached. Hence, developing a reasoning engine to generate goal-driven explanations regarding IA's situational awareness is highly required. To address this challenge, this paper proposes a fuzzy-cognitive-map-based reasoning engine to disclose inferred situations behind IA's executed action. Carla simulator was used to conduct an experimental test in a collaborative driving context. As the results, goal-driven explanations can be generated with a better performance than the baseline method. This work is important to reduce road incidents involving autonomous cars.