Robotic Manipulation (Spatial) on LIBERO-Spatial 1.0 (test)
96.7SRAutoHorizon
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AutoHorizonBackbone=GR00T N1.52026.02 | 96.7 | — | |
| Static OracleExecution Horizon (e)=2, Backbone=GR00T N1.52026.02 | 96 | — | |
| Static Oracle+Backbone=GR00T N1.52026.02 | 96 | — | |
| Static OracleExecution Horizon (e)=8, Backbone=GR00T N1.52026.02 | 95.3 | — | |
| Static OracleExecution Horizon (e)=4, Backbone=GR00T N1.52026.02 | 94.7 | — | |
| Static OracleExecution Horizon (e)=16, Backbone=GR00T N1.52026.02 | 94 | — | |
| RandomBackbone=GR00T N1.52026.02 | 93.3 | — | |
| Static OracleExecution Horizon (e)=1, Backbone=GR00T N1.52026.02 | 92.7 | — | |
| Static OracleExecution Horizon (e)=12, Backbone=GR00T N1.52026.02 | 91.3 | — | |
| NoneVLA Model=DP2026.01 | 90.1 | — | |
| TrojanRobotVLA Model=π02026.01 | 90.1 | 88.1 | |
| State BackdoorVLA Model=OpenVLA2026.01 | 89.5 | 90.1 | |
| TrojanRobotVLA Model=DP2026.01 | 89.2 | 86.2 | |
| State BackdoorVLA Model=ACT2026.01 | 88.2 | 82.4 | |
| TrojanRobotVLA Model=SmolVLA2026.01 | 87.5 | 80.4 | |
| NoneVLA Model=SmolVLA2026.01 | 87.4 | — | |
| NoneVLA Model=π02026.01 | 86 | — | |
| BadVLAVLA Model=DP2026.01 | 85.5 | 84.1 | |
| State BackdoorVLA Model=DP2026.01 | 85.1 | 86.3 | |
| BadVLAVLA Model=OpenVLA2026.01 | 84.6 | 85.4 | |
| NoneVLA Model=ACT2026.01 | 84.3 | — | |
| BadVLAVLA Model=SmolVLA2026.01 | 83.7 | 86.6 | |
| NoneVLA Model=OpenVLA2026.01 | 82.7 | — | |
| TrojanRobotVLA Model=ACT2026.01 | 82.5 | 81.3 | |
| BadVLAVLA Model=ACT2026.01 | 82.4 | 83.8 | |
| TrojanRobotVLA Model=OpenVLA2026.01 | 82.4 | 82.3 | |
| State BackdoorVLA Model=π02026.01 | 82.2 | 81.5 | |
| BadVLAVLA Model=π02026.01 | 81.3 | 83.2 | |
| State BackdoorVLA Model=SmolVLA2026.01 | 80.8 | 86.1 |