Skill registry

360° environment scan with open-vocab object detection. Builds a current semantic view of nearby objects and free space.

Finds and tracks a specific object from natural-language attributes using 3D pose estimation.

Finds and tracks a person using skeleton detection and re-identification across camera views.

Plans and follows person-aware, collision-free indoor routes with real-time replanning.

Moves to a manipulation-ready position relative to a detected surface or object.

Selects and executes a force-limited grasp using visual servoing and grasp pose estimation.

Verifies stable grip after grasp by monitoring force sensors and slip detection.

Maintains a stable whole-body posture while transporting an object at safe velocity.

Gently places a held object on a target surface with force feedback confirmation.

Force-limited transfer to a tracked recipient with grip release on pull detection.

Immediately freezes motion and selects the lowest-risk holding posture. Zero-latency command.

Moves to a stable rest position suitable for extended idle or error recovery.

Alerts the operator for manual intervention when autonomous recovery is not possible.

Opens common cabinet handles using visual servoing and force feedback control.

Directs head cameras at a specified target for detailed inspection or tracking.