Development of an algorithm for the automatic evaluation of Adaptive Cruise Control performance based on subjective and objective data fusion

The purpose of this thesis is the definition of the Key Performance Indicators (KPIs) related to the Adaptive Cruise Control feature. In particular, the scope of this research is to analyse vehicles of different car manufacturers in order to determine which are the performance indicators that are most influential in the subjective evaluations given by different drivers about the Adaptive Cruise Control, and to develop an algorithm which is able to automatically identify the indicators values and evaluate the system during the development phase. The starting point of this work was the analysis of physical and electrical quantities obtained directly through the acquisition of vehicles' internal signals or through the use of an Inertial Measurement Unit (IMU) during test activities performed by qualified drivers for the compilation of the Quality Profile, a document used by the drivers to evaluate the Adaptive Cruise Control feature through various entries. Combining these subjective evaluations with the objective measurements, it was possible to define ranges of values for the KPIs considered as ideal targets for Stellantis vehicles. In order to obtain the values related to the KPIs, a list of test manoeuvres was created, from which it is possible to acquire all the necessary data for the system evaluation. The next step was the development of an algorithm in Matlab that is able to automatically analyse the log files collected during the test manoeuvres and to fill an Excel file with all the values of the KPIs. Moreover, through a further analysis of the subjective evaluations assessed by the drivers and the objective measurements collected during the tests, it was possible to train the algorithm to automatically evaluate the Adaptive Cruise Control system and to compile the Quality Profile relying on the values of the indicators previously identified. In terms of results, the definition of a series of indicators related to Adaptive Cruise Control performance has allowed to move from a totally subjective approach, easily prone to errors and variability, to an objective approach which takes into account the evaluations of qualified drivers. Moreover, the development of a detailed list of test manoeuvres to collect the indicators values has standardised the procedures for the tests execution, and has enabled the comparison between data collected from the test vehicles and those collected from other vehicles of the same car manufacturer or of competitors. The creation of the algorithm has drastically improved the development phase of the Adaptive Cruise Control system on Stellantis vehicles, saving time and money for testing activities and data analysis. The capability of the algorithm to determine the values of the KPIs and to assign evaluations to the entries of the Quality Profile allows to maintain a strong coherence between the measurements collected during the analysis and the evaluations given by the qualified drivers. This research is also helpful for future applications because the algorithm can be modified and trained to perform the analysis of other ADAS features (e.g. Lateral Control, Intelligent Speed Assistance, Predictive Adaptive Cruise Control).