Robotic Tabletop Manipulation on RoboCasa GR1 Tabletop Tasks
58.3Average Success RateABot-M0
Evaluation Results
| Method | Links | |||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ABot-M02026.02 | 58.3 | 86 | 74 | 48 | 46 | 50 | 66 | 70 | 58 | 76 | 66 | 38 | 52 | 66 | 60 | 26 | 54 | 48 | 66 | 64 | 54 | 68 | 64 | 60 | 38 | |
| LDA-1BVis. Rep.=DINO, MMDiT=true, VLM=Qwen3vl2026.02 | 55.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DualCoT-VLArollouts per task=50, training=jointly on all 24 tasks2026.03 | 55.1 | 66 | 64 | 46 | 58 | 30 | 38 | 44 | 54 | 80 | 64 | 46 | 48 | 58 | 74 | 26 | 50 | 56 | 70 | 76 | 52 | 64 | 70 | 60 | 28 | |
| AugVLA-3DNumber of demonstrations=1002026.02 | 54 | 83 | 81 | 46 | 63 | 35 | 62 | 65 | 33 | 71 | 60 | 28 | 52 | 39 | 41 | 25 | 43 | 39 | 47 | 87 | 60 | 67 | 65 | 52 | 31 | |
| GR00T-EI30kVis. Rep.=N/A, MMDiT=false, VLM=Qwen3vl2026.02 | 51.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LDA-0.5BVis. Rep.=DINO, MMDiT=true, VLM=Qwen3vl2026.02 | 50.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gr00TNumber of demonstrations=1002026.02 | 50 | 81 | 77 | 44 | 59 | 27 | 54 | 62 | 30 | 66 | 58 | 24 | 46 | 39 | 37 | 22 | 42 | 38 | 48 | 85 | 56 | 69 | 60 | 49 | 28 | |
| LDA(DiT)Vis. Rep.=DINO, MMDiT=false, VLM=Qwen3vl2026.02 | 48.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3OFTBackbone=Qwen32026.02 | 48.8 | 30 | 76 | 44 | 44 | 32 | 36 | 50 | 40 | 70 | 54 | 38 | 32 | 58 | 52 | 24 | 60 | 50 | 66 | 68 | 44 | 56 | 62 | 54 | 30 | |
| Qwen3OFTrollouts per task=50, training=jointly on all 24 tasks2026.03 | 48.8 | 30 | 76 | 44 | 44 | 32 | 36 | 50 | 40 | 70 | 54 | 38 | 32 | 58 | 52 | 24 | 60 | 50 | 66 | 68 | 44 | 56 | 62 | 54 | 30 | |
| GR00T-N1.5rollouts per task=50, training=jointly on all 24 tasks2026.03 | 48.2 | 54 | 50 | 38 | 60 | 32 | 38 | 38 | 46 | 58 | 62 | 28 | 30 | 60 | 56 | 36 | 52 | 48 | 60 | 52 | 32 | 58 | 44 | 60 | 64 | |
| Qwen3GR00TBackbone=Qwen32026.02 | 47.8 | 46 | 80 | 54 | 48 | 28 | 46 | 48 | 40 | 68 | 52 | 56 | 42 | 44 | 48 | 18 | 60 | 50 | 54 | 70 | 38 | 56 | 50 | 36 | 16 | |
| StarVLAVis. Rep.=N/A, MMDiT=false, VLM=Qwen3vl2026.02 | 47.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3GR00Trollouts per task=50, training=jointly on all 24 tasks2026.03 | 47.8 | 46 | 80 | 54 | 48 | 28 | 46 | 48 | 40 | 68 | 52 | 56 | 42 | 44 | 48 | 18 | 60 | 50 | 54 | 70 | 38 | 56 | 50 | 36 | 16 | |
| GR00T-N1.62026.02 | 47.6 | 51.5 | 13 | 8.5 | 14 | 41.5 | 16.5 | 58 | 46.5 | 68.5 | 65 | 46.5 | 58.5 | 57.5 | 63 | 28.5 | 57 | 43.5 | 51 | 78.7 | 51.5 | 71 | 64.5 | 57 | 31.5 | |
| GR00T-N1.6Vis. Rep.=N/A, MMDiT=false, VLM=Cosmos2026.02 | 47.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GR00T-N1.6rollouts per task=50, training=jointly on all 24 tasks2026.03 | 47.6 | 51.5 | 13 | 8.5 | 14 | 41.5 | 16.5 | 58 | 46.5 | 68.5 | 65 | 46.5 | 58.5 | 57.5 | 63 | 28.5 | 57 | 43.5 | 51 | 78.7 | 51.5 | 71 | 64.5 | 57 | 31.5 | |
| AugVLA-3DNumber of demonstrations=302026.02 | 46 | 60 | 75 | 34 | 45 | 35 | 60 | 46 | 36 | 70 | 57 | 18 | 49 | 34 | 38 | 18 | 40 | 39 | 38 | 72 | 56 | 60 | 55 | 36 | 22 | |
| Qwen3PIBackbone=Qwen32026.02 | 43.9 | 26 | 62 | 42 | 50 | 42 | 32 | 40 | 46 | 60 | 40 | 44 | 44 | 52 | 50 | 28 | 52 | 40 | 36 | 48 | 34 | 64 | 44 | 50 | 28 | |
| Qwen3PIrollouts per task=50, training=jointly on all 24 tasks2026.03 | 43.9 | 26 | 62 | 42 | 50 | 42 | 32 | 40 | 46 | 60 | 40 | 44 | 44 | 52 | 50 | 28 | 52 | 40 | 36 | 48 | 34 | 64 | 44 | 50 | 28 | |
| Gr00TNumber of demonstrations=302026.02 | 43 | 61 | 78 | 36 | 42 | 30 | 58 | 43 | 31 | 68 | 59 | 14 | 50 | 31 | 33 | 12 | 41 | 34 | 35 | 73 | 51 | 54 | 52 | 33 | 18 | |
| Qwen3FASTBackbone=Qwen32026.02 | 39 | 38 | 44 | 56 | 44 | 14 | 14 | 54 | 42 | 58 | 58 | 40 | 36 | 38 | 42 | 18 | 52 | 30 | 48 | 50 | 28 | 34 | 46 | 36 | 16 | |
| Qwen3FASTrollouts per task=50, training=jointly on all 24 tasks2026.03 | 39 | 38 | 44 | 56 | 44 | 14 | 14 | 54 | 42 | 58 | 58 | 40 | 36 | 38 | 42 | 18 | 52 | 30 | 48 | 50 | 28 | 34 | 46 | 36 | 16 | |
| Diffusion PolicyNumber of demonstrations=1002026.02 | 33 | 63 | 75 | 32 | 41 | 30 | 56 | 42 | 16 | 48 | 37 | 14 | 24 | 19 | 24 | 6 | 19 | 14 | 18 | 39 | 37 | 41 | 48 | 34 | 7 | |
| Diffusion PolicyNumber of demonstrations=302026.02 | 21 | 40 | 48 | 25 | 37 | 18 | 43 | 20 | 12 | 28 | 23 | 14 | 16 | 15 | 16 | 7 | 16 | 13 | 14 | 26 | 22 | 27 | 22 | 13 | 2 | |
| UWM(MM-DiT)Vis. Rep.=VAE, MMDiT=true, VLM=Qwen3vl2026.02 | 20 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UWM-1BVis. Rep.=VAE, MMDiT=false, VLM=Qwen3vl2026.02 | 19.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UWM-0.1BVis. Rep.=VAE, MMDiT=false, VLM=N/A2026.02 | 14.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |