Mobile Manipulation on SetTable
90.7Open Fridge Success RateAC-DiT
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AC-DiT2026.02 | 90.7 | — | — | 81.3 | — | 97.3 | 55.6 | 33.3 | 33.3 | 36 | 17.3 | |
| InCoM2026.02 | 87.3 | — | — | 88.9 | — | 100 | 83.8 | 59.4 | 84.1 | 84.1 | 82.5 | |
| ACT2026.02 | 74.6 | — | — | 48.1 | — | 91.5 | 38.2 | 1.6 | 21.2 | 9 | 21.7 | |
| DSPv22026.02 | 73.4 | — | — | 29.9 | — | 98.4 | 50.8 | 1.4 | 65.2 | 1.4 | 85.7 | |
| SG-VLAInput Modalities=Multiview + Depth, Visual Encoder=DINOv2 + SigLIP, LLM Backbone=Qwen2.5-0.5B [32]2026.03 | 67 | 16 | 56 | 36 | 83 | 100 | 60 | — | — | — | — | |
| DP2026.02 | 63 | — | — | 5.3 | — | 89.4 | 39.8 | 0.5 | 54.5 | 2.1 | 63.5 | |
| Base VLM + MultiviewInput Modalities=Multiview, Visual Encoder=DINOv2 + SigLIP, LLM Backbone=Qwen2.5-0.5B [32]2026.03 | 60 | 6 | 53 | 30 | 63 | 93 | 52 | — | — | — | — | |
| Base VLM + Multiview + Depth + HistoryInput Modalities=Multiview + Depth + History, Visual Encoder=DINOv2 + SigLIP, LLM Backbone=Qwen2.5-0.5B [32]2026.03 | 57 | 0 | 47 | 40 | 47 | 100 | 49 | — | — | — | — | |
| OpenVLA + Multiview + DepthInput Modalities=Multiview + Depth2026.03 | 43 | 12 | 41 | 30 | 0 | 67 | 32 | — | — | — | — | |
| WB-VIMA2026.02 | 27 | — | — | 5.3 | — | 87.3 | 34.4 | 1.6 | 57.7 | 1.6 | 60.3 | |
| OpenVLA + MultiviewInput Modalities=Multiview2026.03 | 14 | 6 | 35 | 38 | 0 | 53 | 24 | — | — | — | — | |
| OpenVLAInput Modalities=Default2026.03 | 2 | 0 | 19 | 0 | 0 | 4 | 4 | — | — | — | — |