robotics
The Handoff Became a Robot Skill of Its Own
BATON explores subtasks separately and repairs the transition state before the next contact-rich skill begins.
Summary
BATON explores subtasks separately and repairs the transition state before the next contact-rich skill begins.
BATON freezes the underlying vision-language-action policy and makes each subtask the unit of test-time exploration. Solutions are stored in language memory, turning an approximately multiplicative whole-task search cost into an additive stage-by-stage process. A verifier checks the scene before invoking a contact skill, while handoff and lookahead transitions restore states a successor can inherit. On RoboMemArena, the authors report task-success and cumulative-success gains of 11.6 and 14.9 percentage points over the compared state of the art; these are benchmark results, not evidence of broad real-world reliability.
Why it matters
BATON explores subtasks separately and repairs the transition state before the next contact-rich skill begins.
Limits and context
- On RoboMemArena, the authors report task-success and cumulative-success gains of 11.6 and 14.9 percentage points over the compared state of the art; these are benchmark results, not evidence of broad real-world reliability.
Key claims
BATON explores subtasks separately and repairs the transition state before the next contact-rich skill begins.
Qualification: On RoboMemArena, the authors report task-success and cumulative-success gains of 11.6 and 14.9 percentage points over the compared state of the art; these are benchmark results, not evidence of broad real-world reliability.
Evidence: source-2026-08-18-003
Sources
- arXiv preprint 2608.16889arXiv · primary research
Corrections
No corrections have been recorded for this story.