Computation of driver safety rating using in-vehicle data recorder: Case study of bangkok public transportation

This paper proposes a novel algorithm for the computation of Driver Safety Rating. The Driver Safety Rating is evaluated by detecting dangerous driving patterns based on vehicles data collected via in-vehicle data recorder. Dangerous driving patterns include rapid acceleration, harsh braking and driving over the speed limit and four triggering events are used in the proposed algorithm. A case study using real world data from public mini vans operating throughout Bangkok is deployed in this work. In addition to the proposed algorithm, a fine tuning technique to adjust the calculation process is introduced. An online survey is used to collect passengers feedback in the form of a score to reflect how they actually feel about the safety aspect of the journey. The proposed algorithm is then finely tuned in order to optimise for the scores derived from passengers feedback. Experimental results reveal that only two of the four triggering events have a significant impact on the overall computation of the Driver Safety Rating for this given set of data.