A novel diver pedal behavior model framework incorporating vehicle and road environment information

Abstract

Driver pedal behavior models can benefit vehicle powertrain control and intelligent driving assistance system designs. This paper proposes a novel way to describe the pedal position signal by decomposing it into a series of actions and a nonlinear stochastic driver pedal behavior model is presented under the input-output hidden Markov model (IOHMM) framework. This framework can incorporate the vehicle and road environment information as inputs. A driver model under this IOHMM framework is identified using human driver data collected on a driving simulator and the model behavior is analyzed in comparison with the human driver data. Closed-loop simulation using the model integrated with a vehicle model and a road environment model shows that the outputs of proposed pedal behavior model have similar patterns to those of human drivers.