Measurement of vehicle acceleration in studies of older drivers from GPS position and OBDII velocity sensors

Abstract

This paper demonstrates the validity of vehicle acceleration/deceleration signals derived from 1Hz sampled GPS position and OBDII velocity sensors through comparison to 40Hz sampled accelerometer measurements. Measurement of driver acceleration and deceleration is important because it is a key measure of driving habits. Ideally, these measurements should not require the cost and complexity of installing dedicated accelerometers for long term studies when alternatives are available. The OBDII interface is built-in and GPS sensors can be easily deployed and both are shown to allow derivation of alternative acceleration signals. The results show a maximum average correlation of 0.810 between the GPS and the accelerometers and 0.808 between the OBDII and the accelerometer. This paper analyzes the effects of noise on each of the derivative difference equations and shows that the Central 2-point formula provides the best noise performance whereas the Central 4-point formula (correlation 0.801) would be expected to provide the best performance in a noise free signal. Forward/Backward 3-point are predicted to have similar performance to Central 2-point in noise free signals but are shown to have poor performance (correlation of 0.667 and 0.687 respectively) in the presence of noise.