rskill-playbook-preflight_reach

rskill-playbook-preflight_reach

A kind: playbook rSkill: a symbolic S2 decision procedure the Reasoner reads, not a neural policy. It carries no weights — the authored PLAYBOOK.md is its runtime.

What this skill does

Before dispatching a manipulation skill, it checks the target is within the robot's reachable workspace. It reads the reasoner's ## ROBOT self-model (locomotion, reach, payload), locates the target via spatial memory / detection, and runs a reach check. If the target is reachable it allows the manipulation; if it is out of reach and a mobile base exists it stages an approach stand pose and re-checks; if the arm is fixed (or the target is over the payload limit) it does not dispatch and hands off to a human. Concrete walkthrough: the mug example in PLAYBOOK.md.

How it works

This playbook is content, not code. When installed, the reasoner injects PLAYBOOK.md into its system prompt and follows the SOP, composing tools it already has (recall_object, resolve_place, query_scene, execute_rskill, emit_prompt). It is role: s2 and is never dispatched through ExecuteSkill. Every motion it triggers is an execute_rskill → Action chunk → C++ safety kernel — the playbook holds no actuation authority (CLAUDE.md §1.1).

Observation → action contract

None. A playbook emits no Action chunks and requires no actuators (actuators_required: [], chunk_size: 1). Its "output" is the sequence of tool calls the reasoner makes while following the SOP, bounded by playbook.max_steps.

How it was authored / Upstream provenance

N/A — a playbook is hand-authored, not trained: it has no weights and no upstream model. Its provenance is the authoring decision record (also linked via paper_url). To change behaviour, edit PLAYBOOK.md and bump version.

Supported robots

Embodiment-agnostic — declares the explicit wildcard embodiment_tags: ["any"] (never an empty list). Gated by capabilities_required (has_vision: true — a real RobotCapabilities flag): the loader filters it out on robots without a camera. Navigation / staging are gated at runtime by the composed tools, not by this playbook's flags.

Sensors required

None directly. The tools it composes declare their own sensor needs.

Manifest summary

• kind: playbook, role: s2, actions: [plan], chunk_size: 1.
• playbook.trigger: about to dispatch a manipulation skill on a target whose reachability is uncertain.
• playbook.done_predicate: the target is confirmed within the robot's reachable workspace (or staged so it is), or the task is handed off.
• playbook.max_steps: 8.

Quick start

from openral_core.schemas import RSkillManifest

m = RSkillManifest.from_yaml("rskills/preflight-reach/rskill.yaml")
assert m.kind == "playbook" and m.playbook is not None
print(m.playbook.trigger)

Reproduction

Packaging-only: the manifest + SOP are validated by tests/unit/test_playbook_rskill_manifest.py. There is no benchmark number to reproduce; the playbook's behaviour is exercised by the reasoner integration tests in later phases.

Evaluation

N/A — no eval/*.json; a playbook produces no benchmarkable policy output.

License

• Code / content: Apache-2.0.
• Weights: none.

See also

• PLAYBOOK.md — the decision procedure itself.