List your company
ResearchFormerly WizWand

Robotic Motion Control (Circle task) on Ball-Circle

903.98RewardPPOL-vanilla
506.2736609.5243712.775816.0257Feb 13, 2026
Updated 8mo ago

Evaluation results

MethodLinks
PPOL-vanillaAttack Type=MC2026.02903.9876.36
PPOL-randomAttack Type=MC2026.02830.6465.3
PPOL-vanillaAttack Type=Random2026.02813.457.16
PPOL-vanillaAttack Type=Natural2026.02807.667.78
PPOL-randomAttack Type=Random2026.02776.291.16
PPOL-randomAttack Type=Natural2026.02764.181.43
TCRL-PPOLAttack Type=MC2026.02754.415.56
PPOL-vanillaAttack Type=MAD2026.02728.0520.8
PPOL-randomAttack Type=MAD2026.02727.9321.36
ADV-PPOL(MC)Attack Type=MC2026.02724.328.58
TCRL-PPOLAttack Type=Worst-TC2026.02709.43.84
TCRL-PPOLAttack Type=Natural2026.02669.340
PPOL-vanillaAttack Type=Worst-TC2026.02630.7375.44
PPOL-randomAttack Type=Worst-TC2026.02600.3172.8
TCRL-PPOLAttack Type=MAD2026.02596.450
ADV-PPOL(MC)Attack Type=Random2026.02566.980.44
ADV-PPOL(MC)Attack Type=Natural2026.02559.140.82
TCRL-PPOLAttack Type=Random2026.02550.30
ADV-PPOL(MC)Attack Type=MAD2026.02543.560.38
ADV-PPOL(MC)Attack Type=Worst-TC2026.02521.5725.32