Research map / Multi-agent systems and games / LLM agents
RoboFactory: Exploring Embodied Agent Collaboration with Compositional Constraints
Yiran Qin, Li Kang, Xiufeng Song, Zhenfei Yin, Xiaohong Liu, Xihui Liu, Ruimao Zhang, Lei Bai
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add spatial constraints to logical-only constraints (ablation) | RoboFactory data generation · Lift Barrier | data generation success rate (%): 80.2 -> 95.2 | Improved |
| add spatial constraints to logical+temporal (ablation) | RoboFactory data generation · Three Robots Stack Cube | data generation success rate (%): 84.2 -> 98.9 | Improved |
| add temporal constraints to logical+spatial (ablation) | RoboFactory data generation · Take Photo | data generation success rate (%): 53.8 -> 88.2 | Improved |
| add temporal constraints to logical-only (ablation) | RoboFactory data generation · Lift Barrier | data generation success rate (%): 80.2 -> 85.4 | Improved |
| add temporal constraints to logical+spatial (ablation) | RoboFactory data generation · Take Photo | average episode length: 325 -> 204 | Improved |
| add temporal constraints to logical+spatial (ablation) | RoboFactory data generation · Lift Barrier | average episode length: 115 -> 80.7 | Improved |
| use separate policies instead of shared policy (global view) (ablation) | Diffusion Policy · Place Food | success rate: 5% -> 17% | Improved |
| use separate policies instead of shared policy (local view) (ablation) | Diffusion Policy · Place Food | success rate: 0% -> 20% | Improved |
| use local view instead of global view (separate policy) (ablation) | Diffusion Policy · Lift Barrier | success rate: 26% -> 58% | Improved |
| use local view instead of global view (shared policy) (ablation) | Diffusion Policy · Lift Barrier | success rate: 49% -> 4% | Hurt |
| use local view instead of global view (separate policy) (ablation) | Diffusion Policy · Place Food | success rate: 17% -> 20% | No clear effect |
| increase demonstrations from 50 to 150 (ablation) | Diffusion Policy · RoboFactory 1-agent tasks average | success rate: 25% -> 49% | Improved |
Closest papers by meaning
- LabUtopia: High-Fidelity Simulation and Hierarchical Benchmark for Scientific Embodied Agents (NeurIPS 2025 · similarity 0.87)
- Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Models (NeurIPS 2026 · similarity 0.87)
- COMBO: Compositional World Models for Embodied Multi-Agent Cooperation (ICLR 2025 · similarity 0.87)
- EmbodiedBench: Comprehensive Benchmarking Multi-modal Large Language Models for Vision-Driven Embodied Agents (ICML 2025 · similarity 0.87)
- Collaborative Tree Search for Enhancing Embodied Multi-Agent Collaboration (CVPR 2025 · similarity 0.86)
- PARTNR: A Benchmark for Planning and Reasoning in Embodied Multi-agent Tasks (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
