A neural network for predicting unintentional lane departures

Abstract

Unintended lane departure accidents are due to driver's inattention, incapacitation, and drowsiness. Lane departure warning systems have been developed to enhance traffic safety by predicting/detecting driving situation and alerting drivers to avoid or mitigate traffic accidents. This paper explores effectiveness of a three-layer perceptron neural network in predicting an unintentional lane departure, which to the best of our knowledge has not been reported in the literature. This study used driver experiment data generated by VIRTTEX, a hydraulically powered 6-degrees-of-freedom moving base driving simulator at Ford Motor Company. The experimental data represented 16 drowsy drivers who drove a simulated 2000 Volvo S80 (three hours per driver), which consisted of a total of 3,508 lane departure occurrences. Two-third of the lane departures were randomly selected to generate training examples for the network (82,040 examples for a 0.2-second prediction horizon and 171,112 for a 0.5-second horizon). The number of hidden neurons as well as the input vehicle variables were optimized experimentally through the training process. The optimized network was then used to predict lane departure by processing the entire driving time series of the 16 drivers one by one after all the training data was removed from the time series. The network made a prediction at each sampling moment of the time series and there were over 6.3 million predictions. The overall recall and precision of the optimized network for the 0.2-second horizon were 99.74% and 99.66%, respectively, which degraded to 99.23% and 85.49%, respectively, when the horizon increased to 0.5 s.