A robust observer based on H∞ filtering with parameter uncertainties combined with Neural Networks for estimation of vehicle roll angle

Abstract

Nowadays, one of the main objectives in road transport is to decrease the number of accident victims. Rollover accidents caused nearly 33% of all deaths from passenger vehicle crashes. Roll Stability Control (RSC) systems prevent vehicles from untripped rollover accidents. The lateral load transfer is the main parameter which is taken into account in the RSC systems. This parameter is related to the roll angle, which can be directly measured from a dual-antenna GPS. Nevertheless, this is a costly technique. For this reason, roll angle has to be estimated. In this paper, a novel observer based on H ∞ filtering in combination with a neural network (NN) for the vehicle roll angle estimation is proposed. The design of this observer is based on four main criteria: to use a simplified vehicle model, to use signals of sensors which are installed onboard in current vehicles, to consider the inaccuracy in the system model and to attenuate the effect of the external disturbances. Experimental results show the effectiveness of the proposed observer..

VEHIOT: Design and Evaluation of an IoT Architecture Based on Low-Cost Devices to Be Embedded in Production Vehicles

“…To quantify the accuracy of the sensor, the norm, RMS (root mean square) and maximum errors have been calculated. The norm error as a function of time is calculated as follows [ 5 ]: where: represents the ground truth data, represents the low-cost sensor data and is the mean value of the ground truth data obtained during the period T .…”

“…Nowadays, many researches are focused on the design of observers in order to estimate these variables using the data provided by sensors that are installed on current vehicles, or by low-cost devices in order to solve this problem. In some studies, the vehicle roll angle is estimated by fusing the data of longitudinal acceleration, lateral acceleration, yaw rate and roll rate [ 4 , 5 ], through the fusion of lateral accelerations and roll rate [ 6 ], fusing the data obtained from an inertial angle sensor and a gyroscope [ 7 ], fusing the information obtained from a six-dimensional IMU [ 8 ] or fusing the information provided by a low-cost GPS with the data provided by a wheel speed sensor, steering angle sensor and a yaw rate sensor [ 9 ]. Regarding these works, the most common sensors used to estimate the vehicle dynamics are accelerometers and gyroscopes, which are integrated in an IMU sensor.…”

“…The observers’ design to estimate the vehicles angles is based on different techniques. These techniques are based on artificial intelligence [ 4 , 10 ], Kalman Filter [ 4 , 8 , 10 ], Bayesian Filters [ 9 ], sliding mode observer [ 6 ] and robust observers [ 5 , 11 , 13 ]. All these observers must work under real-time constraints.…”

Vehicle roll stability control with active roll-resistant electro-hydraulic suspension

“…According to statistics from the US National Highway Traffic Safety Administration, rollover crashes contribute 33% to the total number of fatal deaths in the USA, although they account for only 3% of all passenger car crashes [1]. Vehicle rollover accidents are among the most common fatal causes of human death [2].…”

Dynamic Characteristics Analysis of Vehicle Incorporating Hydraulically Interconnected Suspension System with Dual Accumulators

“…As a consequence, research studies on suspension system aiming at improving handling performance and reducing rollover propensity has found prosperity in terms of both active/semiactive controlled and passive suspensions [1][2][3][4][5][6][7][8]. Many active/semiactive control strategies, such as H ∞ control strategies and sliding-mode control strategies, have been utilized onto vehicle to enhance handling performance and ride comfort [9][10][11][12][13][14][15][16]. However, they are too expensive for ordinary passenger cars.…”