Robotic Mosaic explores a material-led workflow for transforming irregular ceramic waste into bespoke mosaic surfaces. Instead of forcing reclaimed tiles into a predetermined system, the process begins by perceiving and digitizing what is available. A vision-guided robotic setup identifies each tile’s shape, color, position, and orientation, creating a live material inventory that can evolve as new tiles are introduced.

Computational algorithms then translate this unpredictable collection into a coherent composition. Using color fields, directional vector fields, surface geometry, and signed distance functions, the system searches for viable positions for each individual tile while respecting its fixed shape and avoiding collisions. The result is not a mosaic generated from standardized digital elements, but a design negotiated between intention and material availability.

The workflow operates across both planar and free-form surfaces. On flat surfaces, automatic packing can be combined with manual decisions available through a Graphical User Interface. For three-dimensional architectural elements, a robotic arm first excavates and compacts an earth formwork. The tiles are then flipped, oriented, and embedded into the shaped earth before concrete is cast. Once the formwork is removed, the tiles become an integral mosaic finish within the concrete surface.

By connecting computer vision, computational design, and robotic fabrication, the project reframes waste as a source of possibility. Irregularity is not eliminated or standardized; it becomes an active part of the design process, enabling discarded materials to generate customized architectural surfaces.