Robotic Fabrication Techniques for Material of Unknown Geometry

Both natural materials such as timber and low-grade or recycled materials are extremely variable in quality and geometry in unprocessed state. Additive digital fabrication processes in robotics in combination with sensor feedback techniques offer large design freedom, high precision and material efficiency and enable a highly customized fabrication and calculation process. Separate studies have been made on scanning, efficient algorithmic arrangement and automated assembly of structures of variable timber elements. In this paper we explore a robotic fabrication process, in which we combine the techniques of scanning, digitally arranging and robotically assembling in one continuous real-time workflow. This means that the final design and appearance only emerge after a unique fabrication process, corresponding to the material used and the assembly sequence. We describe techniques for the simulation modelling and performance analysis using particle simulation, and demonstrate the feasibility through the realisation of the envelope of a robotically assembled double-story timber structure with hand-split wood plates of varying dimensions. We discuss a future use of natural, low-grade or waste material in the building industry through robotic processes. We conclude by analysing the integration of qualitative analysis, physical simulation and the degree of variability of input material and resulting complexity in the computation and fabrication process.