CADENA: Stepwise CAD Reverse Engineering

📅 2026-08-01
📈 Citations: 0
Influential: 0
📄 PDF
🤖 AI Summary
This work addresses a key limitation in existing CAD reverse engineering approaches, which typically generate complete programs in a single pass without intermediate geometric validation, hindering the efficient reconstruction of editable, parametric models. The authors propose a novel paradigm that incrementally constructs CAD programs by adding one modeling operation at a time, guided by a geometric feedback mechanism that compares the current reconstruction against the target shape at each step—mimicking how human engineers model features sequentially. By integrating sequence generation with geometric alignment, this method pioneers geometry-informed, step-by-step reverse engineering and introduces CADENA-Bench, the first benchmark tailored for mechanical parts. Experiments demonstrate significant performance gains over state-of-the-art methods across multiple datasets, including CADENA-Bench, DeepCAD, Fusion 360, and MCB.
📝 Abstract
Computer-Aided Design (CAD) underpins modern engineering, yet converting existing shapes into editable models still demands substantial expert effort. Most AI systems emit the entire CAD program in a single pass, never inspecting the intermediate geometry. In contrast, human engineers build a part feature by feature, checking after each operation what remains to be modeled. We introduce CADENA (Spanish for "chain"), a model that reconstructs a 3D mesh as a parametric CAD program, growing its sequence of operations one at a time and comparing the target with the currently predicted geometry at every step. We also address the lack of benchmarks for evaluating reverse-engineering methods on mechanical parts, introducing CADENA-Bench, a benchmark that measures performance across categories of mechanical parts. CADENA outperforms prior methods on CADENA-Bench and on the DeepCAD, Fusion 360, and MCB datasets. Code is available at https://github.com/zhemdi/cadena, model weights at https://huggingface.co/kulibinai/cadena, and CADENA-Bench at https://huggingface.co/datasets/kulibinai/cadena-bench.
Problem

Research questions and friction points this paper is trying to address.

CAD reverse engineering
3D mesh to CAD
parametric modeling
mechanical parts
editable CAD models
Innovation

Methods, ideas, or system contributions that make the work stand out.

stepwise CAD reconstruction
parametric CAD program
geometry-aware modeling
reverse engineering benchmark
iterative CAD generation