I’ve been thinking about ways to design objects physically with something tangible. LEGO seems like a likely candidate due to its ubiquity and familiarity across all ages and it allows a huge range of objects to be built out of a relatively small library of parts. This extensibility and reconfigurablility makes LEGO look favourable for the basis of a construction kit for design.
However, there are a handful of limitations that Duncan and I explored in an ICED paper. These range from low fidelity and resolution issues to limited materials and build directions. One further limitation that I am considering is how LEGO models are built – they tend to be built up as ‘voxels’ from the bottom up. Not only does this take a considerably long time (3-4hrs for a scale model of a Mini Cooper) for larger models, but more importantly means that if something needs redesigning the model has to be un-built to that point, changed and then rebuilt.
This elemental break down of an object is not intuitive to designers or non-technical users and requires skill to master. Instead, we look at features, such as holes, ribs, corners, surfaces etc, to understand an object and when designing, it is features that are changed.
One method I’m going to explore is to modularise 3D printed models through feature preservation to allow redesign of parts with out having to reprint the whole model again. The inspiration from LEGO is that the parts would be connected (similar to the stud/hole in LEGO) to make the model whole after redesign and printing. The benefit of this is two-fold: 1) it will compress design iterations thorough reducing print times (or parallelising printing) and 2) it will allow for an easy approach to iteration comparison (swap parts in and out) and provide a full physical history of the model. The feature identification and preservation is a strong application for machine learning – something that I’m keen to learn to use. Watch this space…