{"ID":2848726,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.25850","arxiv_id":"2510.25850","title":"Debate2Create: Robot Co-design via Multi-Agent LLM Debate","abstract":"We introduce Debate2Create (D2C), a multi-agent LLM framework that formulates robot co-design as structured, iterative debate grounded in physics-based evaluation. A design agent and control agent engage in a thesis-antithesis-synthesis loop, while pluralistic LLM judges provide multi-objective feedback to steer exploration. Across five MuJoCo locomotion benchmarks, D2C achieves up to $3.2\\times$ the default Ant score and $\\sim9\\times$ on Swimmer, outperforming prior LLM-based methods and black-box optimization. Iterative debate yields 18--35% gains over compute-matched zero-shot generation, and D2C-generated rewards transfer to default morphologies in 4/5 tasks. Our results demonstrate that structured multi-agent debate offers an effective alternative to hand-designed objectives for joint morphology-reward optimization.","short_abstract":"We introduce Debate2Create (D2C), a multi-agent LLM framework that formulates robot co-design as structured, iterative debate grounded in physics-based evaluation. A design agent and control agent engage in a thesis-antithesis-synthesis loop, while pluralistic LLM judges provide multi-objective feedback to steer explor...","url_abs":"https://arxiv.org/abs/2510.25850","url_pdf":"https://arxiv.org/pdf/2510.25850v2","authors":"[\"Kevin Qiu\",\"Marek Cygan\"]","published":"2025-10-29T18:00:16Z","proceeding":"cs.RO","tasks":"[\"cs.RO\",\"cs.LG\",\"cs.MA\"]","methods":"[\"Large Language Model\",\"LoRA\"]","has_code":false}