Vehicle driving behavior predicting and judging using LSTM and statistics methods

Autonomous driving is one of the three major innovations in automotive industry. Autonomous vehicles promise to revolutionize human mobility and vehicle safety. This promise can't be realized without the ability to constantly make the right decisions even in the complex situations. This paper proposed a new decision-making system including a new way using the long short-term memory neural network to predict the states of the vehicles nearby in the short future using the history of which got from the cognitive ability. Based on the future states predicted by the neural network, this paper also proposed some statistics methods to give a classification criterion to judge a vehicle is dangerous, attentive or safe.