Momentum Exchange Attitude Control

Definition

Momentum-exchange attitude control stabilizes the base attitude by deliberately swapping
angular momentum between the base and the manipulator’s links through the joint actuators,
rather than by expending reaction-mass actuators (here, gas-jet thrusters). In
rutkovskii2023control the manipulator itself plays the
role of a momentum-exchange device: torques $M_{\alpha }$ from the electromechanical link drives
react against the base through the dynamic_coupling inertia, producing the
body-stabilizing torque while the gas-jet “working fluid” is conserved (its electrical energy is
recoverable). This is a free-FLYING regime — the base is fully actuated by thrusters — but the
scheme is precisely a propellant-saving alternative to thruster-only attitude control; it differs
from the free-FLOATING regime, where the base is uncontrolled and arm reactions are an unavoidable
disturbance rather than a deliberate control authority. Because the joints have finite travel
($|{\alpha }_i|\le {\alpha }_{i,\max }$), the exchanged momentum eventually saturates the links, requiring
a thruster-driven “unloading mode” — the direct analogue of reaction-wheel desaturation
(see momentum_dumping).

Key Equations

Symbols per notation.md. This source is planar and uses source-faithful symbols
not in notation.md: $\vartheta$ = base (body) attitude angle in the inertial frame; ${\alpha }_i$ =
link joint angles; $M_{\vartheta }$ = gas-jet body torque; $M_{{\alpha }_i}$ = joint actuator torque.
Flag: $\vartheta$ here is a scalar body angle and must not be confused with $\eta$
(quaternion scalar part) in notation.md; $\alpha$ here is a joint angle, distinct from the
CVaR confidence level $\alpha$ in notation.md.

The coupled planar dynamics couple body and joints through the off-diagonal inertia blocks
$A_{12}=A_{21}$, which is the channel the exchange exploits (rutkovskii2023control Eq. 1):

$A(q)\ddot{q}=M(q,u)+F(q,\dot{q}),q=(q_K,q_{\alpha }{)}^{\top }.$

Stabilizing the body angle $\vartheta$ using only a link torque $M_{\alpha 1}$ (no thruster
firing) yields the closed-loop attitude response (Eqs. 5–7), whose stiffness/damping are set by the
reduced efficiency coefficient ${\bar{k}}_{\alpha }({\alpha }_1,\lambda )>0$ — the momentum-coupling gain
from joint torque onto the body:

+ k_A \tilde k_0\,\bar k_\alpha(\alpha_1,\lambda)\,\vartheta = \bar M_\Sigma^d(\alpha_1,\lambda,t).$$ When a link reaches its travel limit, the **unloading mode** restores it ($\alpha_i\to\alpha_i^*$) under thruster torque $M_\vartheta\le M_\vartheta^{\max}$ while holding $|\vartheta|\le\vartheta_{\min}$ (Eq. 10) — the desaturation step that returns control authority to the exchange device. ## Source Support - [rutkovskii2023control](../sources/rutkovskii2023control.md) — primary and sole source: proposes using manipulator "motion exchange" with the base to stabilize attitude and save gas-jet propellant; derives the coupled planar model (Eq. 1), the link-torque attitude loop (Eqs. 5–9), and the thruster-driven manipulator-unloading mode (Eq. 10) plus its stability/limit-cycle analysis. ## Related Topics - [momentum_conservation](momentum_conservation.md) — the conservation law that makes base–arm momentum *exchange* possible; here the base is actuated, so total momentum is not conserved, but the coupling that conservation describes is the same channel the exchange uses. - [angular_momentum_conservation](angular_momentum_conservation.md) — the angular specialization governing how a joint slew reacts onto the base attitude $\vartheta$. - [reaction_null_space](reaction_null_space.md) — the *complement* of this technique: RNS chooses joint motions that produce **zero** base reaction, whereas momentum-exchange control deliberately maximizes (and steers) that reaction to stabilize the base. - [base_disturbance_rejection](base_disturbance_rejection.md) — momentum exchange is one actuation route for rejecting base attitude disturbances $\bar M_\Sigma^d$ without thruster fuel. - [momentum_dumping](momentum_dumping.md) — the unloading mode is exactly momentum dumping: when the arm (acting as the momentum store) saturates at its joint limits, thrusters desaturate it. ## Open Questions - This source is planar (single body angle $\vartheta$, in-plane 3-link arm) and the "exchange" is body $\leftrightarrow$ joint *motion* exchange. Does the propellant-saving benefit and stability guarantee survive a full 3-D attitude (SO(3)) with a redundant arm and a fully-actuated 6-DOF base, where the coupling block becomes $\boldsymbol M_{bm}$ (see [notation.md](../notation.md))? - The unloading frequency is set by joint-limit saturation; for our 6-DOF inspection task, can the arm's redundancy ([reaction_null_space](reaction_null_space.md) self-motion) be used to *postpone* saturation and reduce thruster unloading events, trading propellant against a pointing/coverage cost? - The scheme assumes known inertia parameters $\lambda$ (used to set $\tilde k_0$ via the stationarity condition, Eq. 8). How sensitive is the attitude loop's stability margin to payload-mass uncertainty, which is large when the transported element's mass exceeds the arm's?