Dynamic Singularity

Definition

A dynamic singularity is a configuration at which the velocity map from joint rates to
end-effector motion loses rank because of the system’s mass and inertia distribution, not
its kinematic structure alone. In a free-floating manipulator the spacecraft base reacts
passively (momentum conservation), so the end-effector map ${\mathbf{J}}^{\ast }$ folds in the
inertia properties; where $\det [{\mathbf{J}}^{\ast }]=0$ the end-effector cannot be commanded in
some inertial direction even though no kinematic (Jacobian-rank) singularity is present
(papadopoulos1993dynamic eq 37). Dynamic singularities are therefore inertia-dependent and,
in the free-floating regime, path-dependent (base attitude is a non-integrable function of
joint history); they reduce to ordinary kinematic singularities only in the limit of an
infinitely massive base. In our free-flying circumcentroidal formulation the inertia-dependent
object is the circumcentroidal Jacobian ${\mathbf{J}}_{{\nu }_e}^{\oplus }$ (which subtracts the
mass-averaged Jacobian ${\bar{\mathbf{J}}}_v$), and the coordinate transform $\mathbf{\Gamma }$
goes singular exactly with it.

Free-flying vs free-floating (load-bearing)

papadopoulos1993dynamic and calzolari2020singularity assume a free-FLOATING base
(uncontrolled, momentum-conserving). Our system is free-FLYING (fully actuated 6-DOF base):
${\mathbf{\omega }}_b$ is a commanded input, the momentum constraint is replaced by base
actuation, and the residual conditioning hazard is the inertia-weighted block
${\mathbf{J}}_{{\nu }_e}^{\oplus }$ inside $\mathbf{\Gamma }$ — see
free_flying_vs_free_floating.

Key Equations

Free-floating end-effector map and its dynamic-singularity condition. With a passive base, the
inertial end-effector twist is a function of the controlled joint rates alone through an
inertia-dependent Jacobian ${\mathbf{J}}^{\ast }$:

$$[{\dot{\mathbf{r}}}_E,{\mathbf{\omega }}_E{]}^T={\mathbf{J}}^{\ast }\dot{\mathbf{q}},\det [{}^0{\mathbf{J}}^{\ast }(\mathbf{q})]=0.$$

(papadopoulos1993dynamic eqs 1, 37) — ${\mathbf{J}}^{\ast }$ and $\det =0$ are the source’s own
free-floating symbols; the rank loss is governed by joint configuration through the inertia, not
by base attitude. As base mass/inertia $\to \infty$, ${\mathbf{J}}^{\ast }\to {\mathbf{J}}_{{\nu }_e}$
and dynamic singularities collapse onto kinematic ones.

Circumcentroidal (free-flying) analogue. The inertia-weighted Jacobian that carries the
analogous hazard in our regime is

$${\mathbf{J}}_{{\nu }_e}^{\oplus }={\mathbf{J}}_{{\nu }_e}-\left[\begin{array}{c}{\mathbf{R}}_{eb}\\ 0\end{array}\right]{\bar{\mathbf{J}}}_v,{\bar{\mathbf{J}}}_v=\frac{1}{m}\sum _{j=0}^n{m}^{(j)}{\mathbf{R}}_{jb}^T{\mathbf{J}}_{v_j},$$

(current_sota eq 2.2 / Giordano eqs 14, 7) — the mass-averaged subtraction ${\bar{\mathbf{J}}}_v$
is what makes ${\mathbf{J}}_{{\nu }_e}^{\oplus }$ (and hence the dynamic singularity) depend on the
mass distribution rather than on ${\mathbf{J}}_{{\nu }_e}$ alone.

Shared-singularity result. ${\mathbf{J}}_{{\nu }_e}^{\oplus }$ is the lower-right block of the
coordinate transform $\mathbf{\Gamma }$, so ${\mathbf{\Gamma }}^{-1}$ exists iff
${\mathbf{J}}_{{\nu }_e}^{\oplus }$ is nonsingular — the two go singular together:

$$\mathbf{\Gamma }=\left[\begin{array}{ccc}{\mathbf{R}}_{cb}& -{\mathbf{R}}_{cb}[{\mathbf{p}}_{bc}{]}^{\wedge }& {\mathbf{R}}_{cb}{\bar{\mathbf{J}}}_v\\ 0& \mathbf{E}& 0\\ 0& {\mathbf{G}}_{{\omega }_b}& {\mathbf{J}}_{{\nu }_e}^{\oplus }\end{array}\right],{\sigma }_G\equiv {\sigma }_{\min }(\mathbf{\Gamma }).$$

(current_sota eq 2.4 / Giordano eq 19) — proximity to a dynamic singularity is monitored by the
scalar ${\sigma }_G$.

Empirical equivalence of the two proxies. The minimum singular value of $\mathbf{\Gamma }$
and that of ${\mathbf{J}}_{{\nu }_e}^{\oplus }$ track each other almost perfectly:

$$\mathrm{Spearman}({\sigma }_{\min }(\mathbf{\Gamma }),{\sigma }_{\min }({\mathbf{J}}_{{\nu }_e}^{\oplus }))\ge 0.9969,$$

(current_sota §6 / final.tex §7) — so ${\sigma }_G$ is a faithful single-scalar dynamic-singularity
indicator, and every conditioning layer ultimately diverges through the shared
$({\mathbf{J}}_{{\nu }_e}^{\oplus }{)}^{-1}$ factor.

Source Support

• alali2024reinforcement
• calzolari2020singularity — singularity maps over the workspace for the free-floating regime (visualizes where dynamic singularities lie).
• dai2022review
• ellery2004engineering
• giordano2018workspace
• holt1994inertial
• jin2017reaction
• khatib1987unified
• misra2017optimal
• nanos2011use
• nanos2015avoiding — methods to plan motions that avoid dynamic singularities in free-floating systems.
• papadopoulos1993dynamic — introduces dynamic singularities and the free-floating Jacobian ${\mathbf{J}}^{\ast }$; defines $\det [{}^0{\mathbf{J}}^{\ast }]=0$ and the PIW/PDW workspace split (FREE-FLOATING).
• sone2016reactionless

Related Topics

• kinematic_singularity — the inertia-independent counterpart; dynamic singularities reduce to these as base inertia $\to \infty$.
• generalized_jacobian — the free-floating GJM whose rank loss is the classical dynamic singularity; our ${\mathbf{J}}_{{\nu }_e}^{\oplus }$ is the free-flying analogue.
• circumcentroidal_motion — defines ${\mathbf{J}}_{{\nu }_e}^{\oplus }$ and the $\oplus$ split that locates the dynamic-singularity hazard in our formulation.
• singularity_robust_inverse — damped-inverse / Tikhonov handling invoked once ${\sigma }_G$ approaches the dynamic singularity.
• path_dependent_workspace — the PIW/PDW partition: dynamic singularities make reachability path-dependent in the free-floating case.
• singularity_geometry (result) — the determinant factorization ($\sin {\theta }_3$) that makes the neutral/elbow configuration singular regardless of link mass or length.
• singularity_threshold_cascade (result) — how proximity ${\sigma }_G$ to this singularity schedules the four-layer conditioning stack.

Open Questions

• Quantify how much active base control (free-flying) shrinks or eliminates the dynamic-singularity locus relative to the free-floating $\det [{\mathbf{J}}^{\ast }]=0$ set.
• Whether the $\ge 0.9969$ Spearman equivalence between ${\sigma }_G$ and ${\sigma }_{\min }({\mathbf{J}}_{{\nu }_e}^{\oplus })$ holds across the full inertia-parameter envelope or is configuration-specific.
• Map the calzolari2020singularity / nanos2015avoiding free-floating singularity-map and avoidance results onto the free-flying circumcentroidal coordinates.