|国家预印本平台
首页|MechaFormer: Sequence Learning for Kinematic Mechanism Design Automation

MechaFormer: Sequence Learning for Kinematic Mechanism Design Automation

MechaFormer: Sequence Learning for Kinematic Mechanism Design Automation

来源:Arxiv_logoArxiv
英文摘要

Designing mechanical mechanisms to trace specific paths is a classic yet notoriously difficult engineering problem, characterized by a vast and complex search space of discrete topologies and continuous parameters. We introduce MechaFormer, a Transformer-based model that tackles this challenge by treating mechanism design as a conditional sequence generation task. Our model learns to translate a target curve into a domain-specific language (DSL) string, simultaneously determining the mechanism's topology and geometric parameters in a single, unified process. MechaFormer significantly outperforms existing baselines, achieving state-of-the-art path-matching accuracy and generating a wide diversity of novel and valid designs. We demonstrate a suite of sampling strategies that can dramatically improve solution quality and offer designers valuable flexibility. Furthermore, we show that the high-quality outputs from MechaFormer serve as excellent starting points for traditional optimizers, creating a hybrid approach that finds superior solutions with remarkable efficiency.

Diana Bolanos、Mohammadmehdi Ataei、Pradeep Kumar Jayaraman

机械设计、机械制图工程设计、工程测绘

Diana Bolanos,Mohammadmehdi Ataei,Pradeep Kumar Jayaraman.MechaFormer: Sequence Learning for Kinematic Mechanism Design Automation[EB/OL].(2025-08-12)[2025-08-24].https://arxiv.org/abs/2508.09005.点此复制

评论