Robotic Sand Packing for Reusable Casting Formworks: Integrating Vision-Based Toolpath Feedback for Geometric Accuracy

Kinetic Sand presents promising potential as a reusable casting formwork material due to its cohesive and moldable properties. However, its semi-fluid behavior and tendency to deform under pressure make achieving high geometric accuracy a challenge. This paper introduces a vision-based 3D scanning and comparison methodology to evaluate and control the material behavior of Kinetic Sand during robotic packing. Building on previous research into robotic sand-packing grammars, the study analyzes the effects of various toolpath strategies and end-effector geometries to identify parameters that minimize material deviation. By integrating iterative feedback through digital scanning, the methodology enables the refinement of toolpaths and supports the production of complex cement panels with near-zero formwork waste. The result is a more controlled and repeatable process for fabricating accurate, seamless cast elements using emergent material behavior as a design driver.