Análise comparativa de algoritmos de clusterização para reconhecimento de objetos em sistemas de radar automotivo

Abstract

The great difficulty of the automotive industry is to create a safe and reliable navigation system for the autonomous vehicle, it is something that involves many steps, among them, the fusion of sensors and vehicular communication networks for the recognition of objects, an alternative to this challenge is the use of clustering algorithms in automotive radar systems. Clustering algorithm can be defined as a Machine Learning technique that involves grouping data points, and it works as follows, given a set of data points, we can use a clustering algorithm to classify each data point into a specific group. In theory, data points that are in the same group should have similar properties and/or features, while data points in different groups should have highly different properties and/or features. Clustering is an unsupervised learning method and is a common technique for analyzing statistical data used in many fields. There are several methods that have been developed for the application of clustering, among them we have ten main methods, Affinity Propagation, Agglomerative Clustering, BIRCH, DBSCAN, KMeans, MiniBatch KMeans, Mean Shift, OPTICS, Spectral Clustering, Mixture of Gaussians. In this work, we will present a comparative analysis of the clustering algorithms, to verify which one has the highest efficiency to be used in an automotive radar system for object recognition, a point of extreme importance, for the construction of autonomous vehicles.