Center-of-mass (CoM) regulation drives the whole-system centroid C to a desired
inertial position pcd (and holds it there) by commanding a net centroidal force fc on the system. Because the maximum reachable workspace of a space manipulator is
centered on the system CoM, fixing C in inertial space pins the workspace over a
stationary target so the end-effector can keep operating without a separate repositioning maneuver
(giordano2018workspace). In Giordano’s triangular (cascade)
formulation the centroidal equation decouples from the angular-momentum and end-effector equations,
so fc regulates C independently of the EE controller.
Regime note. The source assumes a fully-actuated base (wb∈R6) but
deliberately uses it in a near-free-floating fashion: base actuators stay off during contact-free
motion (zero fuel) and fire only to absorb contact-induced drift and restore C. For our
free-flying system the base is actuated throughout, so CoM regulation is a continuously-active
coordinated-control task (one branch of coordinated_control), not a
fuel-saving contact-recovery measure.
The centroidal velocity is the linear momentum scaled by total mass, and after the
energy-preserving coordinate transform the CoM channel is an inertially-decoupled double integrator
(Giordano Eq. 9, 13a):
vc=m1p,mv˙c=fc.
The regulation law is a PD (stiffness/damping) controller on the centroidal error x~c=pc−pcd (Giordano Eq. 19; the source
writes this error p~c):
\qquad \boldsymbol{K}_c,\boldsymbol{D}_c \succ 0 .$$
Through the dual of the coordinate transform $\boldsymbol{\Gamma}$ (Giordano Eq. 11), this
centroidal force is realized **only** by the base force $\boldsymbol{f}_b$ (plus a base-torque term
correcting the $\mathcal{B}$-to-$\mathcal{C}$ lever arm): the base actuators are devoted to the
outer CoM/attitude loop and never counterbalance the internal end-effector wrench — the source's
fuel-saving "triangular" actuation split.
## Source Support
- [giordano2018workspace](../sources/giordano2018workspace.md) — primary. Derives the triangular
CoM / angular-momentum / EE dynamics, the PD CoM regulation law (Eq. 19), and the triangular
actuation map; validated on the DLR OOS-Sim (7-DOF arm on a 6-DOF base). Shows that without CoM
control, repeated contacts drift the CoM, stretch the arm, and collapse manipulability into a
practical singularity around $t\approx140$ s; with CoM regulation the workspace and
manipulability are restored after every contact.
## Related Topics
- [center_of_mass_jacobian](center_of_mass_jacobian.md) — the mass-averaged map
($\bar{\boldsymbol J}_v$ / centroidal Jacobian) relating joint and base rates to $\boldsymbol{v}_c$;
it is the kinematic object that CoM regulation acts through.
- [coordinated_control](coordinated_control.md) — CoM regulation is the centroidal/outer branch of
the base–arm coordinated controller; the cascade lets it run independently of the EE task.
- [circumcentroidal_motion](circumcentroidal_motion.md) — the complementary EE-about-$\mathcal{C}$
channel. In giordano2018workspace the $\boldsymbol{\Gamma}$ transform (Eq. 10) decouples the CoM
from the angular-momentum and *internal*-EE ($\boldsymbol{\nu}_{e,int}=\boldsymbol{J}_m^*\dot{\boldsymbol q}$)
dynamics; the circumcentroidal $\oplus$ reformulation (Giordano 2019, not cited here) is the closely
related change of velocity coordinates.
- [dynamic_coupling](dynamic_coupling.md) — arm motion shifts the CoM and disturbs the base;
regulating $\mathcal{C}$ explicitly manages that coupling instead of leaving it to react freely.
- [base_disturbance_minimization](base_disturbance_minimization.md) — the free-floating
counterpart: there the goal is *zero reaction* on the base (reaction null-space); here the
actuated base is used to *hold the CoM*, a different objective for a different regime.
Also relevant: [momentum_dumping](momentum_dumping.md) (the angular-momentum branch of the same
cascade) and [manipulability_measure](manipulability_measure.md) (the workspace-quality scalar that
CoM regulation protects).
## Open Questions
- The source keeps the base near-floating (actuators off except at contact) for fuel economy. For
our free-flying system the base is continuously actuated — does the triangular decoupling and its
stability argument still hold when $\boldsymbol{f}_c$ is active at all times, or does the always-on
base force interact with the EE loop in ways the contact-only analysis does not cover?
- The controller "shares the singularity properties of the transposed-Jacobian free-floating
controller": near a dynamic singularity of $\boldsymbol{J}_m^*$ it loses actuation in the singular
direction rather than failing numerically. How does CoM regulation compose with our
[singularity_robust_inverse](singularity_robust_inverse.md) / damped-inverse handling of the
circumcentroidal Jacobian?
- CoM and momentum feedback are reconstructed from the inertia model plus arm kinematics. How
sensitive is regulation accuracy to inertial-parameter and CoM-location uncertainty, which the
nominal analysis assumes away?
</content>
</invoke>