AI

CADFather: Autonomous CAD Reconstruction through Coordinated Tool Use

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 08/10/2026, 09:18:24 UTC.

Briefings

CADFather is an autonomous agentic system designed to reconstruct editable CAD models from 3D meshes by coordinating complementary tools. It uses a vision-language assistant to inspect renders and intermediate reconstructions, deciding which candidate CAD programs to extend, which tools to invoke, and when to finish. The system learns and algorithmic tools propose operations, while numerical optimization refines parameters. Evaluated on DeepCAD, Fusion360, MCB, CADENA-Bench, CADBench, and BenchCAD, it maintains alternative candidates and preserves the best valid result throughout reconstruction. The project uses pretrained generation and assistant models without additional training. Computational cost and the trade-off between cost and quality are analyzed.

Key points

• CADFather is an autonomous agentic system that coordinates complementary tools to recover parametric CAD programs from 3D meshes.
• The system evaluates reconstruction quality and execution validity on the full DeepCAD, Fusion360, and MCB test sets, as well as on CADENA-Bench, CADBench, and BenchCAD.

Context

Reconstructing an editable CAD model from a 3D shape remains a challenging engineering task. Existing methods can propose CAD operations, but no single source of proposals works equally well across different part geometries and stages of reconstruction. This finding matters as it provides a unified approach for autonomous CAD reconstruction.

Limitations

Evaluation is limited to specific test sets including DeepCAD, Fusion360, MCB, CADENA-Bench, CADBench, and BenchCAD. The system uses pretrained generation and assistant models without additional training, which may limit generalizability to other architectures or training regimes. Uncertainty exists regarding the robustness of the agent's decisions across different part geometries and stages of reconstruction.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 06/11/2026, 18:18:19 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeihwcn3vz7atn53hbxwitoooei56qg2voryhcqhkw57dekhiqdfu2m