Status: draft (agent-authored, 2026-06-22) · Scope: the questionable terms on
generated/terminology_Confusing.md· Verdict authority: the user clarifies at the source — this ledger only traces, it does not rename.
The vocabulary sheet terminology_Confusing.md
carries the standing rule that is the origin of the project-wide
guidance against ghost-chasing:
NEVER make up words to simplify or clarify things for the user. If a term is not clearly scientific (eg. ghost, oracle, lottery), explain where it comes from.
This document discharges that rule for the terms flagged as exotic,
so the vocabulary is settled before it leaks into a
proof. It is the provenance backing for the posture-proof line of work —
posture_proof.md, the
posture_oracle_gap result page, and dynamics_modifications_7dof.md
— and is not itself a proof.
Method. For each term I grepped raw/
(immutable sources), wiki/, and generated/;
checked raw/references.bib for
a corpus citation; and, where the real home is another field, looked up
the canonical external reference via Semantic Scholar (verified DOIs in
the appendix). External references are documented, not added to
references.bib (a deliberate scope decision — they
are non-aerospace and the user will cite at the source).
| verdict | meaning | what to do |
|---|---|---|
| sourced-concept | the word is a metaphor/coinage, but the thing it names is a standard, citeable object | keep the word, gloss it once, cite the concept |
| coined-word (external home) | the word is borrowed from a different field where it is a precise technical term | keep with a one-line gloss; cite the external home if a reviewer presses |
| methodology-nickname | a bespoke name for a test/debugging tactic; no external citation is owed | keep as informal lab vocabulary; never put it in a theorem statement |
| not-a-term | colloquial English masquerading as jargon | do not chase; replace with the specific object it refers to |
| term | one-line gloss | verdict | provenance (corpus bibkey / external candidate) | recommendation |
|---|---|---|---|---|
| ghost (ghost motion) | uncontrolled self-motion component along the coordinated-map null space | sourced-concept | self-motion manifold — Burdick 1989 (ext.); null-space/reactionless
family in corpus: reaction_null_space,
sone2016reactionless,
nenchev2013reaction |
keep as nickname; always gloss as “self-motion” in formal text |
| ghost speedometer (v_n) | scalar reading “how much of v is self-motion” | methodology-nickname | the user’s own instrument: z_a · v |
informal only; call it “self-motion magnitude” in writing |
| the ruler (z_a) | mass-weighted covector that reads 1 on self-motion, 0 on task motion | sourced-concept | dynamically-consistent metric — Khatib khatib1995inertial,
khatib1987unified |
call it “the dynamically-consistent measuring covector” |
| route lottery / arm route | which joint path the redundancy resolution picks depends on null-space policy + history | sourced-concept | path-dependent workspace (PIW/PDW) — papadopoulos1993dynamic;
topic path_dependent_workspace |
keep “route”; “lottery” is fine as a hazard metaphor with a gloss |
| oracle (schedule-κ oracle) | offline kinematics-only predictor of where the demand grazes a singular wall | coined-word (external home) | “test oracle” / “pseudo-oracle” — Davis & Weyuker 1981, Weyuker
1982 (ext.); generic use in akella2024risk |
gloss as “offline kinematic feasibility predictor”; reserve “oracle” for prose |
| witness (witness route q_O) | a constructive trajectory that proves a wall-clear solution exists | coined-word (external home) | witness / certificate of an existential claim — math-logic & complexity (ext.) | keep; gloss as “constructive existence certificate” |
| fiber | the affine solution set Γ⁻¹(y) = x_p + ker Γ for one task command | coined-word (external home) | fiber bundle — geometric mechanics; already cited
(Marsden–Montgomery) in dynamics_modifications_7dof.md
§5.2 |
best-grounded of the lot — keep; cite geometric mechanics |
| section | the rule picking one point on every fiber (a reconstruction convention) | coined-word (external home) | section of a fiber bundle; “mechanical connection” — Kelly & Murray 1995 (ext.) | keep; gloss as “a section (one choice per fiber)” |
| M-orthogonal / dynamically consistent | mass-weighted orthogonality choice for the section | coined-word (external home) | Khatib “dynamically consistent” khatib1995inertial;
“mechanical connection” (ext.) |
already correct terminology — un-hedged |
| frozen-roll oracle | clamp the added roll joint → 7-DOF model must equal 6-DOF exactly; any diff is a bug | methodology-nickname | a test oracle via model reduction / metamorphic relation — Chen et al., metamorphic testing (ext.) | keep as test name; gloss “reduction/differential test oracle” |
| freeze fingerprint | freeze code is inert above σ-floor, so divergence pins a freeze-config difference | methodology-nickname | bespoke differential-provenance test | informal only; “divergence-provenance test” in prose |
| freeze / frozen (kernel freeze, hold-last) | below the floor, reuse the last good inverse/kernel instead of recomputing garbage | sourced-concept | hold-last layer of singularity-robust inversion — chiaverini1997singularity;
registry term in wiki/terminology.md |
descriptive and fine; keep |
| rule | (various: “the section rule”, “adoption rule”, “design rule”) | not-a-term | colloquial English; the example the anti-ghost-chasing rule names | do not chase; in formal text name the specific object |
ghost / ghost motion. The word is your lab
metaphor; the object is the self-motion of a redundant
mechanism — the joint+base motion that leaves everything the mission
watches (EE pose, CoM, base attitude) unchanged. The canonical reference
for self-motion as a manifold structure is Burdick
(1989) (ext.); in our free-flying corpus the nearest sourced
relatives are the reaction null space (nenchev2013reaction,
jin2017reaction)
and reactionless motion (sone2016reactionless).
One subtlety your own terminology note already nails: our “ghost” is the
null space of the coordinated map Γ (arm + a
deliberately actuated base counter-shuffle),
not the free-floating reactionless mode and
not the fixed-base joint null space. So “ghost” denotes
a specific self-motion; keep the nickname, but in a theorem
write “the self-motion of Γ (ker Γ)”. The instrument-metaphors
ghost speedometer (v_n) and the ruler
(z_a) are your own constructions; z_a is rigorously the
dynamically-consistent measuring covector (Khatib), so it earns
a real name even if “ruler” stays in the margins.
route lottery / arm route. The vivid part
(“lottery”: a ±2 % retiming reshuffles the singular encounters without
moving any wall) is a metaphor, but the phenomenon is exactly the
path-dependent workspace (PIW/PDW) of papadopoulos1993dynamic
— see topic path_dependent_workspace.
Keep “route”; gloss “lottery” as “PDW route-sensitivity”.
freeze / frozen. Holding the last good value rather
than recomputing through an ill-conditioned inverse is standard
singularity-robust practice (the hold-last tier alongside DLS in chiaverini1997singularity);
it is already a registry term in wiki/terminology.md.
Descriptive, sourced, fine.
oracle (schedule-κ oracle, frozen-roll oracle). This is the one most worth knowing: “oracle” has a precise home in software testing — a test oracle is the mechanism that decides whether a program’s output is correct, and Davis & Weyuker (1981) literally coin “pseudo-oracle” for programs without a closed-form correct answer (followed by Weyuker 1982, “On Testing Non-Testable Programs”). Your schedule-κ oracle (an offline kinematic predictor of singular bands) and frozen-roll oracle (a reduction test) are both test oracles in exactly this sense. (The other lineage is the CS oracle machine — a black box answering a hard question — which is the colloquial connotation.) Recommendation: in formal writing call the schedule-κ object an “offline kinematic feasibility predictor” and footnote the test-oracle lineage; the word “oracle” is defensible, just not self-explanatory.
witness (witness route q_O). “Witness” (a.k.a.
certificate) is standard in mathematical logic and
complexity theory: an object that demonstrates an
existential statement and can be checked directly (the NP certificate is
the famous case). Your q_O(s) is precisely a witness to “a
wall-clear posture schedule exists” — the constructive half of the
existence argument. Keep it; gloss once as “constructive existence
certificate”. This is good, honest vocabulary, not a coinage.
fiber / section / mechanical connection. The
best-sourced family. The redundant solution set
Γ⁻¹(y) = x_p + ker Γ is an affine fiber; a
reconstruction rule that picks one point per fiber is a
section; the mass-weighted (“M-orthogonal”,
dynamically consistent) choice is the mechanical
connection of geometric mechanics. Your own dynamics_modifications_7dof.md
§5.2 already attributes this to Marsden–Montgomery and
Khatib; Kelly & Murray (1995),
“Geometric phases and robotic locomotion,” is the cleanest robotics-side
citation (ext.). This is rigorous applied differential geometry — keep
all of it.
frozen-roll oracle and freeze fingerprint are test/debugging tactics, not concepts in a theorem. The frozen-roll trick is a metamorphic / differential test (clamp roll=0 ⇒ 7-DOF ≡ 6-DOF is a known input/output relation that serves as the oracle; cf. metamorphic testing, Chen et al., ext.). The freeze fingerprint exploits a code-inertia invariant to localize a config difference — a bespoke differential-provenance test with no external owner. Keep both as lab vocabulary; never let them into a formal statement.
rule. Across generated/ and
raw/, “rule” is colloquial English (“the section rule”,
“adoption rule”, “design rule for λ”). It is not a reified object in
control theory and needs no provenance — it is, in fact, the
kind of word the anti-ghost-chasing rule warns about. In formal text,
replace it with the specific object: “the section (reconstruction
convention)”, “the operating-point adoption criterion”, “the damping
schedule”, etc.
naumann2020probabilistic
uses “ghost objects” — but that means radar false
positives (phantom detections from wave reflections) in
autonomous-driving sensor fusion. That is a homonym,
unrelated to your “ghost motion” (a null-space mechanical mode). Citing
Naumann as the provenance of “ghost” would be wrong; the real anchor is
the self-motion / reaction-null-space literature above. Flagged so it
never makes it into a citation by accident.
s;
code/concept name. step-jump — logged smoothness metric
(per-step ‖Δν_e‖); defined in your terminology table. trusted
baseline — the provenance-verified reference run. These are
code names; no external citation is owed.generalized_jacobian,
kinematic_singularity,
and the singularity sources). No provenance gap.references.bib)Documented as leads for you to cite at the source; out of scope for
the aerospace .bib.
| concept | candidate reference | verified id |
|---|---|---|
| self-motion manifold (“ghost”) | Burdick, J. (1989), “On the inverse kinematics of redundant manipulators: characterization of the self-motion manifolds,” Proc. IEEE ICRA, 264–270 | DBLP conf/icra/Burdick89 |
| test oracle / pseudo-oracle (“oracle”) | Davis, M. & Weyuker, E. (1981), “Pseudo-oracles for non-testable programs,” Proc. ACM ’81, 254–257; Weyuker, E. (1982), “On Testing Non-Testable Programs,” Comput. J. 25(4):465–470 | DOI 10.1145/800175.809889; DOI 10.1093/comjnl/25.4.465 |
| metamorphic testing (“frozen-roll oracle”) | Chen, T.Y., Cheung, S.C., Yiu, S.M. (1998/2020), “Metamorphic Testing: A New Approach for Generating Next Test Cases” | arXiv 2002.12543 |
| fiber bundle / mechanical connection (“fiber”, “section”, “M-orthogonal”) | Kelly, S. & Murray, R. (1995), “Geometric phases and robotic locomotion,” J. Robotic Systems 12(6):417–431; Marsden–Montgomery (geometric mechanics) | DOI 10.1002/rob.4620120607 |
| witness / certificate (“witness”) | standard in mathematical logic & complexity (NP certificate); e.g. Arora & Barak, Computational Complexity | — (textbook concept) |
When you do pick up posture_proof.md, the
vocabulary is now safe to use with a one-line gloss
each, in roughly descending order of how defensible the word is
to a hostile reviewer:
A concise version of these verdicts is mirrored in wiki/INSIGHTS.md.