Mobile Manipulation on Floorplans 7 - 10 rooms (test)
100Success Rate (SR)Always Interact
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Always InteractEpisode Horizon (g)=102026.02 | 100 | 1.16 | 2,565.52 | 0.2248 | |
| Clean + S/PEpisode Horizon (g)=102026.02 | 100 | 1 | 3,865.1 | 0.1443 | |
| Always InteractEpisode Horizon (g)=152026.02 | 100 | 1.15 | 3,321.25 | 0.1771 | |
| Clean + S/PEpisode Horizon (g)=152026.02 | 100 | 1 | 4,693.99 | 0.1324 | |
| LLM-driven constraint-based planning frameworkEpisode Horizon (g)=20, Environment Knowledge=known2026.02 | 100 | 1.23 | 2,435.48 | 0.1574 | |
| Always InteractEpisode Horizon (g)=202026.02 | 100 | 1.36 | 3,974.88 | 0.0943 | |
| Clean + S/PEpisode Horizon (g)=202026.02 | 100 | 1 | 5,169.58 | 0.0841 | |
| LLM-driven constraint-based planning frameworkEpisode Horizon (g)=15, Environment Knowledge=known2026.02 | 94.14 | 1.57 | 2,356.94 | 0.191 | |
| LLM-driven constraint-based planning frameworkEpisode Horizon (g)=10, Environment Knowledge=known2026.02 | 90.61 | 1.54 | 1,594.36 | 0.2591 | |
| Always DetourEpisode Horizon (g)=152026.02 | 66.26 | 3.21 | 1,094.18 | 0.1373 | |
| Always DetourEpisode Horizon (g)=102026.02 | 66.06 | 3.23 | 758.76 | 0.1963 | |
| LLM-driven constraint-based planning frameworkEpisode Horizon (g)=10, Environment Knowledge=unknown2026.02 | 65.45 | 3 | 2,270.42 | 0.0837 | |
| LLM-driven constraint-based planning frameworkEpisode Horizon (g)=15, Environment Knowledge=unknown2026.02 | 64.85 | 2.88 | 2,843.94 | 0.0632 | |
| Always DetourEpisode Horizon (g)=202026.02 | 64.09 | 3.43 | 1,381.61 | 0.0661 | |
| LLM-driven constraint-based planning frameworkEpisode Horizon (g)=20, Environment Knowledge=unknown2026.02 | 62.27 | 2.79 | 4,497.8 | 0.0344 |