AI-assisted, curvature-aware post-slicing G-code optimization for material-efficient FDM printing

Abstract Fused Deposition Modeling (FDM) slicers typically apply uniform process parameters across entire parts, leading to inefficient material usage and limited adaptation to local geometric or functional requirements. This work presents a post-slicing G-code optimization framework that operates directly at the instruction level without modifying the original CAD model or re-running the slicer. The framework integrates three modules: (1) curvature-aware adaptive layer height control, (2) function-specific reinforcement for load-bearing regions, and (3) AI-based weak-zone detection using unsupervised clustering. Physically correct extrusion is enforced through incremental ΔE handling in relative mode (M83). The method is validated through three case studies: a quadcopter frame, an ASTM D638 tensile test specimen, and a complex mounting bracket, all printed using a Creality Ender-3 printer (0.4 mm nozzle, Nylon filament) with baseline G-code generated by Ultimaker Cura 5.0. Compared to the original toolpaths, the optimized G-code preserves geometric fidelity (RMS deviation < 0.05 mm) while reducing effective material extrusion by approximately 22–25% across the three geometries. Print time increased by 7–9% due to feedrate smoothing and localized reinforcement. These results suggest that instruction-level, post-slicing optimization can improve material efficiency for functional FDM components across diverse geometries, though further validation across additional printers, materials, and part complexities is needed to establish full generalizability.