An Expert System for Evaluation Driver Behavior Based on Fuzzy Fusion of Smartphone Sensors

Monitoring and evaluating driving behavior indirectly reduces accidents and fuel consumption. But this evaluation is not available due to the need for expensive equipment. In this study, an expert system is presented which allows fuzzy evaluation of drivers` behaviors by using smartphone sensors. The proposed model, first identifies different types of maneuvers such as changing lane, road ramps, turning left or right based on cell phone sensors fusion. Then, the expert system uses fuzzy C-mean clustering technique to determine the overall behavior of the drivers into two categories, aggressive and safe, based on the type of maneuver and the lateral acceleration during the maneuver. Results show fusion of smartphone inertial measurement sensor (IMU) sensors based on the adaptive neuro fuzzy inference system (ANFIS) detect correctly the type of maneuvers near 96%.  Also, in order to validate the results of assessing driver behavior, the well-known Driver Anger Scale questionnaire is used. The output obtained from the proposed model confirms the results of the questionnaire.