TO-Agents: A Multi-Agent AI Pipeline for Preference-Guided Topology Optimization

Topology optimization can generate efficient structures, but designers often must manually translate qualitative intent, such as desired visual style, product experience, or manufacturability into solver settings that are not directly tied to those preferences. We present TO-Agents, a multi-agent AI framework that connects natural-language design intent with iterative topology optimization. The framework converts a human-provided problem description into validated solver inputs, runs a topology optimization solver, renders the resulting 3D topology, and uses multi-view vision-language reasoning with an independent judge agent to critique each result and revise solver parameters. We evaluate the framework on two long-horizon design tasks: a cantilever beam benchmark and a phone-stand product design. In both tasks, the designer specifies an aesthetic preference for hierarchically branched structures inspired by natural tree morphologies, and the system performs four revision cycles across ten independent replicates. TO-Agents produces at least one preference-aligned design in 60% of trials for each case study, corresponding to up to 6x more successful trials than an ablated pipeline without visual or historical feedback. Judge scores and human evaluations show that the pipeline can identify effective parameter levers, recover from poor revisions, and expand design exploration. A manufacturing agent further post-processes top-ranked designs for additive manufacturing, enabling end-to-end intent-to-prototype design. We also identify failure modes, including overshooting, selective memory, misplaced tools, and incorrect parameter reasoning. These results suggest that agentic topology optimization can shift designers from low-level parameter tuning toward higher-level specification of form and function, while highlighting safeguards needed for reliable autonomous engineering design.

📋 한 줄 요약

**[Topology Optimization / Multi-Agent AI]** TO-Agents가 자연어 design intent를 solver input으로 변환·multi-view VLM judge로 비평·매개변수 수정 — 두 task에서 60% 성공률·ablation 대비 최대 6× 향상, 4 revision cycle·10 replicate로 end-to-end intent→prototype.

🎯 핵심 기여도

Topology optimization이 효율적 구조 생성 가능하지만 디자이너가 시각 스타일·제품 경험·제조성 등 정성적 의도를 solver 설정으로 수동 번역해야 하며 이 설정이 선호와 직접 연결되지 않음을 지적.
TO-Agents 제시 — 자연어 design intent와 iterative topology optimization을 연결하는 multi-agent AI 프레임워크.
인간이 제공한 problem description을 검증된 solver input으로 변환, topology optimization solver 실행, 결과 3D topology 렌더링, multi-view vision-language reasoning과 독립 judge agent로 각 결과 비평·solver 파라미터 수정.
Long-horizon design task 2개(cantilever beam 벤치·phone-stand 제품) 평가 — 디자이너가 nature-inspired hierarchically branched 구조 aesthetic 선호 지정, 4 revision cycle·10 independent replicate 수행.

💡 핵심 아이디어

Topology optimization을 디자이너의 정성적 의도와 직접 연결하려면 LLM·VLM 기반 multi-agent가 자연어→solver input 변환, multi-view 시각 비평, parameter 수정 cycle을 자동 수행해야 하며, judge agent의 visual·historical feedback이 단순 ablated pipeline 대비 6×까지 성공률을 끌어올린다.

🔬 기술적 접근법

**방법론**: TO-Agents — natural language→solver input→TO→3D render→VLM judge→parameter revise 다중 agent 파이프라인.
**핵심 기법**: (1) 자연어 problem description을 검증된 solver input으로 변환, (2) Topology optimization solver 실행, (3) 결과 3D topology 다각 뷰 렌더링, (4) Multi-view VLM reasoning + 독립 judge agent로 비평, (5) 4 revision cycle·10 replicate로 parameter 수정·manufacturing agent로 additive manufacturing 후처리.

📊 주요 결과

Cantilever beam·phone-stand 양 task에서 60% trial이 preference-aligned 디자인 생성.
Visual·historical feedback 없는 ablated pipeline 대비 최대 6× 성공률.
Judge score·human evaluation으로 효과적 parameter lever 식별·poor revision 회복·design exploration 확장 확인.
Manufacturing agent로 end-to-end intent-to-prototype 가능.
실패 모드 식별: overshooting, selective memory, misplaced tools, incorrect parameter reasoning.

💭 의의 및 한계

**의의**: 디자이너를 저수준 parameter tuning에서 고수준 form·function 명세로 이동, agentic engineering design 자율성 첫 step, additive manufacturing까지 end-to-end. **한계**: 60% 성공률로 여전히 신뢰성 보강 필요, 4 가지 실패 모드 명시, 자율 engineering 설계의 safeguard 추가 필요.

🚀 실용적 활용

AI 보조 generative engineering design.
Aesthetic·기능 동시 만족 구조 자동 생성.
제품·건축 design exploration 자동화.