Home Knowledge Base the Distilled Reading Heuristic (DRH

the Distilled Reading Heuristic (DRH

No mentions found

This entity hasn't been tracked yet, or Iris is still building its knowledge base.

Related Articles from SNS

When Video Misreads: Closed-Loop Distillation of Reading Heuristics for Exploratory Manipulation Trace QA

Announce Type: new Abstract: Exploratory manipulation often turns an apparent failed attempt into the key evidence for what to do next. For example, a robot pulls a locked cabinet drawer, fails, and only succeeds after opening the lock. The failed pull reveals a latent precondition (the drawer is locked) that determines the minimal-success action chain (the fewest actions that complete the task), here [lock-open, drawer-pull].

arXiv CS 1d ago