n this paper we tailor the dead-zone based mechanism presented in [3] to the well-known kinematic observer for the estimation of vehicle lateral velocity. We extend the previous results on the dead-zone observer to linear parameter varying systems. The proposed mechanism maintains the structure of the kinematic observer but inserts an adaptive dead-zone at the output injection term. This dead-zone mechanism partially 'cuts' the noise and increases the noise rejection performance allowing for the selection of a larger observer gain. We use this freedom to increase the observer gain to attenuate constant bias errors in the acceleration measurements. The proposed solution is easy to implement and requires only measurements acquired from standard on-board sensors. The adaptation parameters are selected solving a suitable Linear Matrix Inequality (LMI), and no manual tuning is required. We show the effectiveness of the proposed solution through numerical simulations.
A kinematic observer with adaptive dead-zone for vehicles lateral velocity estimation / De Pascali, L; Biral, F; Cocetti, M; Zaccarian, L; Tarbouriech, S. - ELETTRONICO. - (2018), pp. 511-516. (Intervento presentato al convegno AMC 2018 tenutosi a Tokyo nel 9 March 2018 through 11 March 2018) [10.1109/AMC.2019.8371146].
A kinematic observer with adaptive dead-zone for vehicles lateral velocity estimation
De Pascali, L;Biral, F;Cocetti, M;Zaccarian, L;Tarbouriech, S
2018-01-01
Abstract
n this paper we tailor the dead-zone based mechanism presented in [3] to the well-known kinematic observer for the estimation of vehicle lateral velocity. We extend the previous results on the dead-zone observer to linear parameter varying systems. The proposed mechanism maintains the structure of the kinematic observer but inserts an adaptive dead-zone at the output injection term. This dead-zone mechanism partially 'cuts' the noise and increases the noise rejection performance allowing for the selection of a larger observer gain. We use this freedom to increase the observer gain to attenuate constant bias errors in the acceleration measurements. The proposed solution is easy to implement and requires only measurements acquired from standard on-board sensors. The adaptation parameters are selected solving a suitable Linear Matrix Inequality (LMI), and no manual tuning is required. We show the effectiveness of the proposed solution through numerical simulations.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione