ML-MAS: A Hybrid AI Framework for Self-Driving Vehicles

Machine Learning (ML) techniques have been shown to be widely successful in environments that require processing a large amount of perception data, such as in fully autonomous self-driving vehicles. Nevertheless, in such a complex domain, ML-only approaches have several limitations. In this paper, we propose a hybrid Artificial Intelligence (AI) framework for fully autonomous self-driving vehicles that uses rule-based agents from symbolic AI to supplement the ML models in their decision-making. Our framework is evaluated using routes from the CARLA simulation environment, and has been shown to improve the driving score of the ML models.