A humanoid pick-and-place policy exhibits initial-pose-dependent hand preference, which can be suppressed or induced by swapping two specific arm joints, and broader initial-pose training data reduces this dependence.
Octo: An open-source generalist robot policy,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Policy-Induced Hand Priors in Humanoid Dual-Arm Manipulation: Diagnosing and Mitigating Initial-Pose Dependence
A humanoid pick-and-place policy exhibits initial-pose-dependent hand preference, which can be suppressed or induced by swapping two specific arm joints, and broader initial-pose training data reduces this dependence.