MobiDriveScore — A system for mobile sensor based driving analysis: A risk assessment model for improving one´s driving

Assessment of driving behavior and estimation of risk therein is an important use case for telematics and vehicular networking technology domain, which has gained interest in the fleet management and consumer verticals, but above all with the car insurance firms. These firms have started incorporating the results of such analysis to provide their customers with "pay as you drive" insurance plans, which decide the initial premium based on the driver\´s past history of crashes and violations. Then the parties enter into an evaluation period of about two months (say), at the end of which premium is recalculated based on analysis of accelerometer, GPS and on-board diagnostic (OBD II) data. However, the consumer has no knowledge of the progress of his or her evaluation due to lack of access to device data from insurance provider. One major criterion of such evaluation is based on the detection and aggregation of high-risk maneuvers by the driver, which typically constitutes of hard bump, sharp cornering, hard stop and speed limit violations. In this paper, we propose a novel method named MobiDriveScore for routine ventures using which the consumer can assess his/her own driving pattern. Thus the consumer can consciously reduce the risk associated with his/her driving by using MobiDriveScore. For MobiDriveScore, detection of maneuvers is done using accelerometer and GPS sensors only, which are integral part of modern day smartphones. Using MobiDriveScore, the user can choose an alternate route or choose to travel a little early to improve upon the premium rates and stay safe. It will also help aggressive drivers to become more passive. Experiments have been conducted and the initial results of MobiDriveScore are found to be encouraging.