robotics
The Robot Chose Actions That Split Its Competing Theories
OHCAM expands conditional action hypotheses only when simpler models stop explaining the observations.
Summary
OHCAM expands conditional action hypotheses only when simpler models stop explaining the observations.
The online learner maintains a belief over possible action models, selects experiments where those hypotheses disagree and grows more complex conditional or quantified effects only when evidence demands them. Across six planning domains, the authors report more solved tasks than baseline methods under limited and noisy observations. Two Kinova Gen3 tasks provide a real-robot check. The experiments support sample-efficient model discovery in the tested domains; they do not establish unrestricted causal learning.
Why it matters
OHCAM expands conditional action hypotheses only when simpler models stop explaining the observations.
Limits and context
- The online learner maintains a belief over possible action models, selects experiments where those hypotheses disagree and grows more complex conditional or quantified effects only when evidence demands them.
- The experiments support sample-efficient model discovery in the tested domains; they do not establish unrestricted causal learning.
Key claims
OHCAM expands conditional action hypotheses only when simpler models stop explaining the observations.
Qualification: The online learner maintains a belief over possible action models, selects experiments where those hypotheses disagree and grows more complex conditional or quantified effects only when evidence demands them.
Evidence: source-2026-09-01-009
Sources
- arXiv preprint 2608.30955arXiv · primary research
Corrections
No corrections have been recorded for this story.