Robotic Task Generation on 900 Generated Tasks Suite
99.1Task ClarityRoboGene
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| RoboGene2026.02 | 99.1 | 98.76 | 98.99 | 63.23 | 91.52 | 98.99 | |
| Human2026.02 | 96.44 | 84.11 | 94.89 | 35.8 | 17.69 | 94.78 | |
| Gemini 2.5 Pro2026.02 | 87.91 | 86.71 | 66.91 | 21.72 | 24.58 | 67.9 | |
| GPT-4o2026.02 | 79.22 | 85.55 | 76.22 | 31.05 | 25.42 | 74.67 | |
| Rule-based2026.02 | 55.33 | 39 | 40.44 | 21.45 | 48.4 | 48.11 |