{"paper":{"title":"AssemblyBench: Physics-Aware Assembly of Complex Industrial Objects","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"AssemblyBench dataset and AssemblyDyno model advance physics-aware assembly planning for industrial objects.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Anoop Cherian, Bernhard Egger, Danrui Li, Jiahao Zhang, Moitreya Chatterjee, Suhas Lohit, Tim K. Marks","submitted_at":"2026-05-13T00:44:09Z","abstract_excerpt":"Assembling objects from parts requires understanding multimodal instructions, linking them to 3D components, and predicting physically plausible 6-DoF motions for each assembly step. Existing datasets focus on simplified scenarios, overlooking shape complexities and assembly trajectories in industrial assemblies. We introduce AssemblyBench, a synthetic dataset of 2,789 industrial objects with multimodal instruction manuals, corresponding 3D part models, and part assembly trajectories. We also propose a transformer-based model, AssemblyDyno, which uses the instructional manual and the 3D shape "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"AssemblyDyno outperforms prior works in both assembly pose estimation and trajectory feasibility, where the latter is evaluated by our physics-based simulations.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The synthetic dataset and physics simulations sufficiently capture the shape complexities and physical constraints of real industrial assembly tasks.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"AssemblyBench dataset and AssemblyDyno transformer model enable physics-aware prediction of assembly sequences and trajectories for complex industrial objects from multimodal instructions and 3D shapes.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"AssemblyBench dataset and AssemblyDyno model advance physics-aware assembly planning for industrial objects.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8275c42eec6565200a058571bdaab48ea4123def78bd96115a3472208c21a75c"},"source":{"id":"2605.12845","kind":"arxiv","version":1},"verdict":{"id":"147799eb-2964-4020-b824-67d8c271ed76","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:37:05.590411Z","strongest_claim":"AssemblyDyno outperforms prior works in both assembly pose estimation and trajectory feasibility, where the latter is evaluated by our physics-based simulations.","one_line_summary":"AssemblyBench dataset and AssemblyDyno transformer model enable physics-aware prediction of assembly sequences and trajectories for complex industrial objects from multimodal instructions and 3D shapes.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The synthetic dataset and physics simulations sufficiently capture the shape complexities and physical constraints of real industrial assembly tasks.","pith_extraction_headline":"AssemblyBench dataset and AssemblyDyno model advance physics-aware assembly planning for industrial objects."},"references":{"count":48,"sample":[{"doi":"","year":2021,"title":"The ikea asm dataset: Understanding people assem- bling furniture through actions, objects and pose","work_id":"e9284f93-742f-4aee-a531-9298e546d4c8","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Qi Charles, Hao Su, Mo Kaichun, and Leonidas J","work_id":"1b9ea2e7-59d1-46fd-abc3-d0377d84f449","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"A point set generation network for 3D object reconstruction from a sin- gle image","work_id":"ee80af9d-b2a5-4b12-a189-ed86388c32d1","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2006,"title":"R. Hadsell, S. Chopra, and Y . LeCun. Dimensionality reduc- tion by learning an invariant mapping. InIEEE Computer Society Conference on Computer Vision and Pattern Recog- nition (CVPR), pages 1735–17","work_id":"8f49669a-d31d-4bbc-8ad7-0fd63686d4fd","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"ProMQA-Assembly: Multimodal Procedural QA Dataset on Assembly","work_id":"2f39c333-ec37-43c2-8462-733d3943e109","ref_index":5,"cited_arxiv_id":"2509.02949","is_internal_anchor":true}],"resolved_work":48,"snapshot_sha256":"e4347d286d4707ad44098b6dd5c089f9d58a4e27bc8d071ef36a62b422b95fdf","internal_anchors":4},"formal_canon":{"evidence_count":2,"snapshot_sha256":"e8f580d1f748c6c7d6292abf42d8dec8525bd5e3f5441ae9d0a35fa09bc7d7d2"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}