Abstract. In familiar design domains, expert designers are able to quickly focus on "good designs", based on constraints they have learned while exploring the design space. This ability to learn novel constraints is a key aspect in which design differs from traditional optimization; the constraints on the search are constantly re-defined based on the search experience itself. Moreover, such constraints are often implicit, i.e. the designer may find it difficult to articulate these constraints and provide reasons for them. Here, we ask if computer-aided-design systems can discover such implicit constraints in well-understood design situations, where the function can be articulated clearly enough to be quantified in terms of performance metrics. By considering function across a large number of design instances, patterns of functional feasibility may be learned as a byproduct of evaluating different designs. We show how patterns of functional infeasibility result in novel constraints that rule out certain regions of the design space. We demonstrate this process using examples from the design of simple locking mechanisms, and as in human experience, we show that the nature of the constraints learned depends on the extent of exposure in the design space, and may be widely variable in early stages. We also show how the process of design change, when the design space is modified, e.g. by adding new design variables, can build on patterns learned on past designs. In conclusion, we discuss the ramifications of this process on chunking and representational change, and also on design creativity.

No abstract

In the widespread endeavour to standardize a vocabulary for design, the semantics for the terms, especially at the detailed levels, are often defined based on the exigencies of the implementation. In human usage, each symbol has a wide range of associations, and any attempt at definition will miss many of these, resulting in brittleness. Human flexibility in symbol usage is possible because our symbols are learned from a vast experience of the world. Here we propose the very first steps towards a process by which CAD systems may acquire symbols is by learning usage patterns or image schemas grounded on experience. Subsequently, more abstract symbols may be derived based on these grounded symbols, which thereby retain the flexibility inherent in a learning system.In many design tasks, the "good designs" lie along regions that can be mapped to lower dimensional surfaces or manifolds, owing to latent interdependencies between the variables. These low-dimensional structures (sometimes called chunks) may constitute the intermediate step between the raw experience and the eventual symbol that arises after these patterns become stabilized through communication. In a multi-functional design scenario, we use a locally linear embedding (LLE) to discover these manifolds, which are compact descriptions for the space of "good designs". We illustrate the approach with a simple 2parameter latch-and-bolt design, and with a 8-parameter universal motor.

We consider two questions related to functional part families: a) how to characterize function in a computational framework, and b)

No abstract