Reclaimed wood presents a valuable resource for circular construction, but its geometric irregularity limits reuse within conventional standardized workflows. This study introduces a vision-based Grasp-to-Mill (GtM) workflow that enables direct robotic processing of irregular reclaimed timber through an integrated perception-to-machining pipeline. Within this pipeline, a step-wise RGB-D reconstruction method rapidly localizes non-standard logs, while a feature-aware strategy formulates 6-DoF grasp selection as a multi-objective optimization for stable, collision-free handling without manual intervention. Subsequently, adaptive milling adjusts tool orientation to follow natural curvature and prevent spindle interference during joint fabrication. Validated on full-scale elements and a 2025 Venice Biennale pavilion, the GtM framework embeds perception and decision-making directly into robotic processing. This demonstrates a scalable approach for upcycling irregular biomass into reliable components for circular construction. • A vision-based robotic workflow enables direct processing of irregular reclaimed wood • Feature-aware grasp planning ensures stable handling of non-standard timber elements • A grasp-to-mill pipeline integrates perception, grasping, and adaptive robotic milling • The workflow reduces energy-intensive pre-processing in reclaimed timber fabrication • Full-scale pavilion validation demonstrates feasibility for circular construction practice
Gao et al. (Wed,) studied this question.