Automated vehicles represent one of the most active research topic in different engineering fields in the recent years; in particular, transportation systems theory, mathematical modeling, informatics and electronics are all involved in the development of more efficient, accurate and error resilient systems regarding this topic. For modeling purpose, it is useful to introduce a vehicle classification, but the current 6 SAE levels are too detailed, without sharp borders in some case, to be useful for practical applications; thus a 2-class meta-classification has been proposed but it seems too coarse. This paper starts from an analysis of the 6 SAE levels and of sensors developed and used in modern vehicles to establish a relationship between them; this study is useful to devise a new effective 3-class meta-classification useful to transport systems analysis involving automated vehicles.