A knowledge-based automation concept, called automated copied driving, has been introduced for a special driving process. Thereby, a database of tool paths for component shapes is employed. New components are produced through the composition of appropriately transformed parts of the data pool. Up to now, building the database has been the main issue due to the complex cataloging of tool paths.This paper presents an automated approach for cataloging that is fast and universally applicable. Therefore, tool paths are parameterized by probabilistic density functions, which, subsequently, are used for tool path derivation. For the computation, a bivariate kernel density estimation is applied.