Analysis of Car-Following Behavior using Dynamic Probabilistic Models Identification of Driving Mode Transition using Dynamic Bayesian Networks

摘要

Dynamic probabilistic models of car-following behavior are constructed by learning dynamic Bayesian networks (DBN), like auto-regressive hidden Markov models (AR-HMM), using data from driving experiments on actual expressway. Car-following behavior is characterized as transition process between several driving modes of dependency on the leading car.