Ground vehicle roll rate and pitch rate signals coming from the inertial sensors usually suffer from the drift problem: the sensor readings are slightly biased with respect to the true values. This unknown drift will have a negative impact on many control applications using these roll and pitch rate signals. This paper presents a method to estimate this drift value while the vehicle is moving, without requiring very accurate vehicle velocity or yaw rate signals. First, a few global roll and pitch angle samples are approximately estimated at some special moments during the vehicle movement. Then a linearized kinematic model is used to formulate a regression problem to estimate the drifts, given the global roll and pitch angle samples. CarSim simulation results show that the proposed method works well in both small and large drift situations.

This content is only available via PDF.
You do not currently have access to this content.