Robotic pouring
Why robots?
Traditional rubble methods rely on a high level of skill, effort, and time. In today’s fast-paced building industry, there is little room for slow, unpredictable processes that require constant adaptation – such as manually working a pile of irregular pieces of mixed material into a careful composition or carving pieces into a perfect fit for construction. However, recent advances in computational design and digital fabrication allow us to disassociate the technique from the substantial manual effort. New technologies such as machine vision and object recognition, algorithms, and robotic fabrication allow customization. Today, we can combine the speed and precision of machines with the flexibility of craftspeople, which opens up a new specter of possibilities for processing reused materials.
Understanding rubble from a digital perspective
When a craftsperson builds a dry-stone wall, they constantly, and often subconsciously, evaluate each stone piece before placing it. Is it too small, does it have the right shape, does it buckle? For a robot to do the same, the material has to be understood from a digital perspective. In the case of dry stone construction, that means that a large quantity of data about the piece is necessary: its geometry, size, and surface quality. Such placing techniques require a lot of information about the individual rubble pieces for construction, ranging from a 2D scan to a full 3D scan of each object. Pouring methods, on the other hand, work with very little information – one only needs to know the quality and the sizes.
Developing robotic pouring and placing
Based on our material and technique studies, we believe that a wide range of rubble techniques will become available through the digitalization of two simple processes: pouring and placing. For example, methods like jammed rubble and casting through robotic pouring, techniques like course preparation for brickwork or dry stacking through robotic placement, and techniques like slip form or Roman concrete through a combination. Within this project, we developed and tested simple workflows for both.
Robotic placing