BenchmarkPAW-GEN-10
Environmente07Place loaded boxes in container (bimanual)
- Episodes
- Pi 0.5 30 Claude Fable 5.2 (D10) 30 Claude Opus 5.5 (D10) 30 DiT-Flow 30 Gemini 4 Argon (D10) 30 GPT-6 Astra (D10) 30 GR00T N1.7 30 MolmoAct 2 30 Rho 30
- Total duration
- Pi 0.5 28:16 Claude Fable 5.2 (D10) 141:44 Claude Opus 5.5 (D10) 42:55 DiT-Flow 6:02 Gemini 4 Argon (D10) 37:09 GPT-6 Astra (D10) 40:07 GR00T N1.7 11:10 MolmoAct 2 9:48 Rho 21:56
- Mean duration
- Pi 0.5 56.5 s Claude Fable 5.2 (D10) 283.5 s Claude Opus 5.5 (D10) 85.8 s DiT-Flow 12.1 s Gemini 4 Argon (D10) 74.3 s GPT-6 Astra (D10) 80.2 s GR00T N1.7 22.3 s MolmoAct 2 19.6 s Rho 43.9 s
- Success
- Pi 0.5 0% Claude Fable 5.2 (D10) 7% Claude Opus 5.5 (D10) 3% DiT-Flow 0% Gemini 4 Argon (D10) 3% GPT-6 Astra (D10) 10% GR00T N1.7 3% MolmoAct 2 0% Rho 3%
- Manipulation density
- Pi 0.5 Share of time manipulating, five definitionsJoints moving99.9%End-effector moving98.3%Gripper engaged59.7%Contact force > 5 N62.6%Gripper engaged or in contact84.3% Claude Fable 5.2 (D10) Share of time manipulating, five definitionsJoints moving73.0%End-effector moving40.9%Gripper engaged74.2%Contact force > 5 N76.3%Gripper engaged or in contact88.2% Claude Opus 5.5 (D10) Share of time manipulating, five definitionsJoints moving93.5%End-effector moving72.6%Gripper engaged69.7%Contact force > 5 N77.1%Gripper engaged or in contact84.5% DiT-Flow Share of time manipulating, five definitionsJoints moving98.3%End-effector moving88.5%Gripper engaged16.5%Contact force > 5 N42.4%Gripper engaged or in contact47.9% Gemini 4 Argon (D10) Share of time manipulating, five definitionsJoints moving90.3%End-effector moving75.6%Gripper engaged56.2%Contact force > 5 N76.2%Gripper engaged or in contact83.5% GPT-6 Astra (D10) Share of time manipulating, five definitionsJoints moving75.0%End-effector moving49.7%Gripper engaged64.2%Contact force > 5 N74.1%Gripper engaged or in contact79.8% GR00T N1.7 Share of time manipulating, five definitionsJoints moving100.0%End-effector moving98.0%Gripper engaged63.2%Contact force > 5 N56.4%Gripper engaged or in contact83.4% MolmoAct 2 Share of time manipulating, five definitionsJoints moving100.0%End-effector moving98.7%Gripper engaged50.8%Contact force > 5 N62.4%Gripper engaged or in contact76.5% Rho Share of time manipulating, five definitionsJoints moving100.0%End-effector moving99.8%Gripper engaged63.4%Contact force > 5 N68.8%Gripper engaged or in contact84.4%
- SPARC smoothness
- Pi 0.5 Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. Claude Fable 5.2 (D10) Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. Claude Opus 5.5 (D10) Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. DiT-Flow Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. Gemini 4 Argon (D10) Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. GPT-6 Astra (D10) Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. GR00T N1.7 Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. MolmoAct 2 Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. Rho Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015.
Episode
Held-out benchmark environment
0:001:03