An approach to the integration of design and manufacturing knowledge is presented that uses a feature-based design environment and an unsupervised learning algorithm to categorize features into a setup for machining. Setup generation is a pivotal step in automated process planning as it greatly influences machine/tool selection, machining sequences and fixture configuration. The proposed algorithm and architecture incorporate multiple objective functions into setup generation. Intersecting and nonintersecting features within a setup are identified and classified using an associative memory. After the feature sequence has been determined, several algorithms are proposed to obtain the best tool sequence for creating the features in a setup. © 1994.