Initiation Safety: A Missing Dimension in Generalist-Robot Safety
For developers of generalist robots, this work highlights a missing safety layer that could prevent unauthorized social actions, though it remains at the proposal stage with no empirical results.
The paper argues that current generalist-robot safety frameworks neglect a critical dimension: whether the robot should initiate a first social action (e.g., greeting, grasping) at all, which they term initiation authorization. They propose a PAS (probe-authorize-speak) framework and demonstrate its feasibility on a doorway humanoid, comparing it with direct-init on logged traces.
Safety for generalist robots is usually discussed in terms of motion or dialogue. We argue a third question is missing: should the robot take its first hard-to-undo social action at all, such as a greeting, an uninvited grasp, or stepping into someone's space? We call this initiation authorization. Current frameworks rarely treat it as a separate safety layer. Today's stacks often skip this step: a high engagement score or a confident VLA rollout is treated as permission to act. But seeing a person is not the same as having their consent to be addressed. We frame initiation authorization within generalist-robot safety and contrast it with post-plan VLA guardrails, implementing PAS (probe-authorize-speak) on a doorway humanoid, comparing it with direct-init on logged traces, and proposing a three-condition user study, with open questions on metrics, governance, and where initiation ends and foundation-model generation begins.