Robots assemble custom wooden designs with less manual tuning
On-Demand Robotic Assembly via Differentiable Geometric Part Repair
Robotics
Summary
Turning a digital design into a physical wooden assembly usually takes months of expert work to fit the parts for robots. The authors created an automated system that starts from user ideas, generates 3D designs, and adjusts part shapes so that robots can assemble them more easily. Their method fixes errors like overlapping parts and adapts components so robots can screw them together correctly. They tested this with two robots successfully building ten different wooden structures from natural language prompts.
What this means in practice
- •For custom furniture makers: Automatically generate and optimize wooden assembly designs that robots can build directly from simple text descriptions.$Commercial implications: Enables on-demand robotic manufacturing of bespoke wooden furniture with minimal expert setup, allowing scalable personalization and rapid production.
- •For robotic automation engineers: Incorporate differentiable geometry repair into robotic assembly workflows to reduce the manual tuning needed for part fit and robot compatibility.
Authors
Millicent Schlafly, Fabio Schaub, Diogo Costa Pais, Luca Lelli, Janne Dvorak, Claire Colmont, Sven Marti, Mark D. Fuge
Abstract
Transitioning from a digital design to a robotic assembly process currently requires months of expert manual tuning to reconcile part geometries with robotic constraints. This paper presents an end-to-end, autonomous pipeline for the design and physical construction of bespoke wooden assemblies. A generative AI agent translates user prompts into initial 3D geometries, balancing the visual fidelity of the design with select physical constraints. The assemblability of the design is further improved by a gradient-based repair stage that backpropagates through a graph attention network surrogate to adjust component geometries. In addition to correcting for disjointed and overlapping components, we demonstrate hardware-specific corrections, differentiably optimizing the geometry of components to enable robot screwdriving for 86.7% of 60 novel natural language inputs, significantly outperforming prior work by a factor of ten. For ten of the structures, we physically demonstrate assemblability with two UR5e robots. This work marks a meaningful step toward on-demand robotic manufacturing, enabling the rapid production of customized, low-volume goods.