A comparison of driving behaviour prediction algorithm using multi-sensory data on a smartphone

The causes of accidents on highway in any countries come from vehicle condition, human error, and highway physical conditions. From the accident statistic on the highway in Thailand, the major cause of the accidents is from aggressive driving behaviour. One way to decrease highway traffic accidents is to provide some information, such as alerts and warnings to the car drivers. Currently, people all around the world use smartphones on a daily basis. Many useful applications can be installed on the smart phones. Thus, this paper presents driver's behaviour detection algorithm using multi-sensory data on a smartphone to detect driving events and provide real-time feedback to drivers while driving. From variation in hardware of device available, the algorithm allows user to customize the algorithm according to the device configuration. Pattern matching of different sensory data was applied to find driving event. The algorithm demonstrates 71 - 80% accuracy in correct prediction of driving events.