Numerical Dataset Analysis and Sensors Architecture Review for Vehicle Dynamics Application

Modern vehicles are nowadays largely equipped with sensors sampling several physical quantities in order to monitor the status of on-board subsystems and more in general give all the information needed on the vehicle dynamics to Electronic Control Units (ECUs). Sensors mounted are featured by different nature as far as the working principles and its techniques of fabrication are concerned. However, it can happen that the actual sensor architecture is not sufficiently robust to prompt the correct behaviour of the vehicle itself. If some physical quantities are not adequately monitored by dedicated sensors, the sampled dataset can be incomplete or not correlated with the real vehicle dynamics. In order to provide for the lack of sensors, it is important to consider both the analysis of the problem via numerical and a review of the whole sensor architecture. This paper aims to analyse the dataset sampled by on-board sensors of the vehicle to evaluate the odometry and estimate the physical quantities not monitored because of the absence of such sensors through a numerical model of the vehicle. The numerical model adopted is fitted and tuned on real tests performed. The information recovered contributes to define the dataset with more accuracy. Furthermore, a schematization of an optimal sensors architecture for vehicle dynamics application is provided to reduce the lack of information on the odometry, considering a frequent critical condition due to the interference given by the weather.