LLM의 행동 데이터 생성과 반복 가중치 개선을 통한 로봇 모델 제작
Building a robot model requires repeated data collection, policy training, failure analysis, and modification. This work places the LLM in the role of model producer and proposes a production loop connecting behavior-data generation to improvement of the trained model's weights. A person initially specifies the desired model and provides feedback from real-world use of the resulting model, while AI performs the intermediate generation, training, modification, and testing. In the first case, RaccoonBot, demonstrations from a tool-using teacher train a geometrically integrated GRU policy, and the LLM designs, implements, and executes correction procedures to produce candidate weights. Numerical updates use AdamW, demonstration-frequency-based distance estimation, and Newton-based CLOSE correction. In stored simulated development evaluations, successes for the selected seed increased from 238/252 to 252/252 unique starts and from 12/24 to 24/24 conditions for seven- to eight-step programs, comparing the existing approach-distance policy with the combined candidate. This limited-task case demonstrates an implementation path, not fully autonomous model production or independent generalization. Follow-up research will compare improvement mechanisms and human intervention under matched budgets and investigate reuse of the production loop across robots and purpose-specific models. Version 0.8.1: structurally rewritten bilingual research preprint, Korean (15 pages) and English (12 pages). This revision clarifies the model-production loop, the LLM's role, numerical update mechanisms, provenance, and the scope of stored simulated development results. It does not report new experiments or establish a world-first claim. Previous version 0.7.1 remains available at https://doi.org/10.5281/zenodo.24212005. Sole author: 유용현. Research repository: https://github.com/dreaminglizard/llm-driven-robot-policy-learning. CC BY 4.0 applies to the papers in this record, not automatically to linked code, datasets, or model weights. SHA256 paper_ko.pdf: c4ed5a5a907f90dae0107913fda237b74d65b3a6af95c88ded4278baf281e2b8. SHA256 paper_en.pdf: aa7ed8ad3c9874c6009baca8be282f361faf115f06b96753b1714f78beb64baf.
Authors
- 용현 유
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.5281/zenodo.24211881
- Primary Topic
- Reinforcement Learning in Robotics
- Type
- preprint