Knowledge Representation for Driving Context Cognition in a Smartphone-based ADAS

In the near future, autonomous vehicles will drive on behalf of the passengers. The vehicle shall be equipped with accessible knowledge whenever and wherever it is needed. Such knowledge includes driving and traffic rules. The idea is simple: the vehicle will access its knowledge-base in order to perceive its environment and recognize its driving situation; it will react accordingly, behaving exactly like a human driver. More particularly, it is capable of detecting collision before it actually happens, and maneuvers to avoid it. The vehicle of the future has its ways of collecting parameters from different sources (sensors, smart city, Internet, etc.) and combine them to perceive the driving situation. An autonomous vehicle should be able to reason out and decide what actions to be implemented. Such action is implemented by sending signals to the vehicle's actuators. In the same manner, a semi-autonomous vehicle would behave the same, except that the driving assistance (notification, alert or danger messages) are sent to a human driver, and the action signals to the actuators. In this work, knowledge representation related to the vehicle is implemented using ontology. SWRL (semantic web rule language) is used to formulate knowledge related to driving and traffic rules. The driving assistance mechanism ADAS (advanced driving assistance system) is deployed in a smartphone. A use case is used to validate the knowledge representation and formulation.