Vehicle velocity estimation using nonlinear observers

Nonlinear Observer for Vehicle Velocity with Friction and Road Bank Angle Adaptation - Validation and Comparison with an Extended Kalman Filter

“…See [12,11,16]. Since these tire slips are functions of the dynamic states and wheel steering angles and wheel speeds, we will often write F i = F i (v x , v y , r, δ i , ω i ; µ H ), where we have added µ H to denote dependence on the road conditions through the maximum tire-road friction coefficient.…”

“…For a more complete description of the vehicle model, we refer to [12,11,14]. Here, we briefly sum up the vehicle model in the horizontal plane aṡ…”

“…This is an overall velocity observer with adaptation of the maximum friction coefficient and the road bank angle. The detailed development of the different parts of the observer is described in [12,11,7,6,10], * Also affiliated with NTNU, Department of Engineering Cybernetics, N-7491 Trondheim, Norway and will be briefly summarized herein. Furthermore, we want to assess, to a certain degree, its performance in terms of accuracy and robustness, tuning, and computational complexity, in comparison with the EKF [19] applied to the same problem.…”

Set-membership LPV model identification of vehicle lateral dynamics

“…Remark 2 -Lateral acceleration measurements can be obtained through low cost accelerometers, while many different effective approaches can be found in the literature to compute a reliable estimate of the vehicle longitudinal velocity (see, e.g., Imsland et al, 2006) using information from on-board sensors commonly available on production cars. Availability of such low cost measurements of the scheduling variables makes it possible to use the proposed LPV models in real-time applications such as, for example, virtual sensors for on-line estimation of sideslip angle and yaw rate.…”

Distributed robust vehicle state estimation

“…A model-based vehicle lateral state estimator is developed in [9] using a yaw rate gyroscope, a forward-looking monocular camera, an a priori map of road superelevation and temporally previewed lane geometry. On the other hand, the kinematic method uses acceleration and the yaw rate measurements from an inertial measurement unit (IMU) and estimates the vehicle velocities employing Kalman-based [10], [11], or nonlinear [12] observers. This method does not employ a tire model, but instead the sensors bias and noise should be identified precisely to have a reliable estimation.…”