Key points are not available for this paper at this time.
Hybrid additive – subtractive manufacturing (HAM) couples layer-wise deposition with CNC finishing. The workflow still lacks an automated way to recover reliable datum planes from warped and defect-laden point cloud scans. Conventional random sample consensus-style (RANSAC) fitting collapses toward tilted ‘average’ planes once systematic warpage or material spill occurs, inflating localisation error and endangering machining. We propose the Six-Directional Orthogonal Datum Plane Identification (SODPI) framework, which integrates AM process physics and design priors into the geometric reasoning process to handle specific manufacturing distortions. SODPI replaces unconstrained greedy search with six axis-aligned hypothesis tests, adds constrained RANSAC plus iterative inlier removal, and merges homotype patches to preserve semantics. On real hybrid additiveFormula: see textsubtractive manufacturing (HAM) coupons and industrial point clouds, SODPI lowers angular error from Formula: see textto Formula: see text, positional error from 0.42 mm to 0.03 mm, and root mean square (RMS) residual from 0.89 mm to 0.12 mm while keeping total stack-up below 1 mm. The method establishes a closed-loop digital thread from scan to work coordinate system, reducing setup time from a duration ∼12mm of manual probing to 90 s of fully automated operation, enabling predictive quality gates and automated feedback to process simulation that bridge digital models and physical manufacturing processes (Formula: see text).
Wang et al. (Mon,) studied this question.