Adaptive Robotic Carving

Abstract

The paper presents the developments of a series of methods to train a fabrication system for the integration of material performances in timber manufacturing processes, combining robotic fabrication together with different sensing strategies and machine learning techniques, and their further application within a prototypical design to manufacturing workflow. The training cycle, spanning from the recording of skilled human experts to autonomous robotic explorations, aims to encapsulate different layers of instrumental knowledge into a design interface, giving designers the opportunity to engage with material and tool affordances as process driver. The training methods are evaluated in a series of experiments and design iterations, proving their potential in the development of customized design to manufacturing workflows and integration of material performances, with a specific focus on timber.