For a free-floating space manipulator (uncontrolled base; arm motion reacts on the spacecraft),
Papadopoulos & Dubowsky partition the reachable workspace into two regions. The Path Independent
Workspace (PIW) is the set of inertial locations that can never be reached in a dynamically
singular configuration — reachable by any path lying within it. The Path Dependent Workspace
(PDW) is the complement: locations that may be reached in a dynamically singular configuration depending on the path the end-effector takes to get there, so they are reachable by some paths but
not others. The phenomenon exists because the spacecraft attitude is not a function of the joint
angles alone but of the history (path) of the motion — a consequence of the non-integrability of
angular momentum — so a given workspace point maps to infinitely many configurations q,
some of which are singular. The PDW/PIW split is a free-floating artifact; it presumes an
uncontrolled base.
A dynamic singularity is a rank loss of the free-floating end-effector Jacobian J∗
(here at fixed base, 0J∗(q); the base-attitude transform diag(T0,T0) is always invertible):
det[0J∗(q)]=0.
The set of singular configurations forms hypersurfaces Qs,i in joint space. Each maps,
via the end-effector’s barycentric distance R=∥rE−rCM∥
from the system centre of mass, to a spherical shell in inertial space; each hypersurface sweeps a
shell-bounded volume (Rmin,i,Rmax,i). The PDW is the union of these volumes and the PIW is
its complement within the reachable workspace WR:
\qquad \text{PIW}=\mathcal W_R \setminus \text{PDW}.$$
(Symbols $R,\,R_{\min,i},\,R_{\max,i},\,\boldsymbol Q_{s,i}$ and the reachable workspace $\mathcal W_R$
are not in notation.md; introduced source-faithfully here. $\boldsymbol J^{*}$ is the registered
Papadopoulos Jacobian — distinct from $\boldsymbol J_g$ and $\boldsymbol J_{\nu_e}^{\oplus}$.)
## Source Support
- [papadopoulos1993dynamic](../sources/papadopoulos1993dynamic.md) — primary and sole source: defines the
PDW/PIW partition, proves the singularities of $\boldsymbol J^{*}$ are path-dependent via the
non-integrability of angular momentum, and gives the planar two-link example (reaching point B from A
fails on the straight path AB but succeeds on the detour ACB). Also notes design levers that shrink the
PDW: large/infinite base inertia, or mounting the arm at the base CoM (planar case eliminates the PDW).
## Related Topics
- [dynamic_singularity](dynamic_singularity.md) — the rank-deficiency event ($\det{}^{0}\boldsymbol J^{*}=0$) whose path-dependent placement in inertial space *creates* the PDW.
- [free_floating_dynamics](free_floating_dynamics.md) — the regime that gives rise to the phenomenon; an actuated base would let attitude track a command rather than drift with the path.
- [angular_momentum_conservation](angular_momentum_conservation.md) — its *non-integrability* is the root cause: attitude depends on path history, so a point maps to many (some singular) configurations.
- [generalized_jacobian](generalized_jacobian.md) — the Umetani GJM ($\boldsymbol J_g$) is the companion free-floating map; $\boldsymbol J^{*}$ here plays the same role and shares the dynamic-singularity property.
- [dynamic_coupling](dynamic_coupling.md) — base reaction to arm motion is the coupling that makes attitude (hence reachability) path-dependent.
- [kinematic_singularity](kinematic_singularity.md) — fixed-base singularities are path-independent (solve $\det\boldsymbol J(\boldsymbol q)=0$); they remain a subset of the dynamic singularities here and bound the reachable workspace.
## Open Questions
- The PDW/PIW partition assumes an **uncontrolled (free-floating) base**. For our **free-flying** system the base attitude is a commanded state, not a path integral of arm motion — does the path-dependence collapse, leaving only the (path-independent) configuration-dependent singularity set, or does a residual coupling-driven analogue survive?
- Papadopoulos shows large base inertia shrinks the PDW; the free-flying analogue would be base actuation authority. Is there a quantitative trade between control bandwidth/torque margin and an effective PIW for an actuated base?
- "The efficient construction of paths to reach points in the PDW is still an open area of research" (source). For an actuated base, does this reduce to standard singularity-avoidance planning, or does coupled base-arm guidance change the path-construction problem?