Research And Development On Smartphone Based Vehicle Driving Behaviours Sensing System

The growth of the number of vehicles has brought us comfort and convenience in our lives but on the other hand the rate of growth of accidents has increased exponentially.The accidents reported have been occurring due to either drivers’ or vulnerable road users’ fault.Useful methods that can be implemented in the vehicle and monitor the drivers’ driving behaviours need to be applied in order to minimize or prevent the occurrence of accidents.Different applications and systems that aim at analyzing steering maneuvering have been developed.Camera-based applications’ performances are seriously constrained by their high visibility requirements.On the other hand,camera free approaches such as V-Sense lack clear explanation on how to effectively detect different steering maneuvers.Considering only one bump for making judgement of the turns either left or right is not enough since the driving style differs from one driver to the other.This research makes several key contributions to addressing some camera free approaches’challenges.Number of bumps detected,horizontal displacement and vehicle’s heading angle are the suggested factors that have assisted in improving the accuracy and driving behaviours of the drivers.The smartphones represented an appropriate platform to work out on the driving behaviours of the drivers through the use of embedded sensors such as GPS,accelerometer and gyroscope.Data obtained from accelerometer and GPS contained a lot of noise and some biases as well.A moving average filter has been adopted so as to remove noise from the raw gyroscope readings.Data have been fused by using a Kalman filter in order to estimate the velocity.Through the sensor fusion method,different driving styles recognition can be achieved.The research has provided useful insights on some of the factors that needed to be taken into consideration in order to improve the accuracy and performance of the existing camera free applications.They assisted in not only improving the classifications’ accuracy of the vehicle steering maneuvers but also showing that can be achieved with a low computational overhead.