Macro Mini Manipulation

Definition

Macro-/mini-manipulation (Khatib 1995) treats the serial combination of two stages — a proximal macro-manipulator ($n_M$ DOF, large reach, slow) and a distal lightweight mini-manipulator ($n_m$ DOF, small range, high bandwidth) — as a single redundant system of $n=n_M+n_m$ joints controlled in operational space. The mini stage is assumed to span the $m$-dimensional task space ($n_m\ge m$, $n_M\ge 1$), so the fast distal links carry out the end-effector task while the slow proximal links reposition the mini stage within its limited joint range. Khatib derives the construct for a fixed (grounded) macro base, but explicitly applies it to a free-flying robotic system (Russakow & Khatib 1992) in which the free-flying base is treated as the macro structure and the manipulator as the relatively lightweight mini structure — the regime of our system (whose base is fully-actuated and 6-DOF; Khatib does not state the Russakow base’s DOF or actuation), treated as a redundant serial chain rather than a momentum-conserving free-floating one.

Key Equations

Symbols per notation.md.

Block decomposition of the combined joint-space inertia ($\mathbf{M}$ here is Khatib’s $A$; macro block first, mini block second), whose mini-stage block ${\mathbf{M}}_{22}$ equals the inertia of the mini-manipulator considered alone (Lemma 1):

$$\mathbf{M}(\mathbf{q})=\left[\begin{array}{cc}{\mathbf{M}}_{11}& {\mathbf{M}}_{12}\\ {\mathbf{M}}_{21}& {\mathbf{M}}_{22}\end{array}\right],{\mathbf{M}}_{22}={\mathbf{M}}_m.$$

Reduced effective inertia (Theorem 2 / Corollary 1): the operational-space (task) inertia of the macro-/mini system is bounded above, in every direction, by that of the mini-manipulator alone — the combined system never feels heavier at the end effector than the light distal stage:

$${\mathbf{w}}^{\top }{\mathbf{\Lambda }}_o\mathbf{w}\le {\mathbf{w}}^{\top }{\mathbf{\Lambda }}_{m(o)}\mathbf{w}\forall \mathbf{w},\mathbf{\Lambda }=(\mathbf{J}{\mathbf{M}}^{-1}{\mathbf{J}}^{\top }{)}^{-1}.$$

Dextrous dynamic coordination (§5.3): operational-space task torques plus a midrange-attraction torque ${\mathbf{\tau }}_0$ injected through the dynamically-consistent null-space projector $\mathbf{N}=\mathbf{E}-{\mathbf{J}}^{\top }{\bar{\mathbf{J}}}^{\top }$, so the macro stage keeps the mini stage off its joint limits without perturbing the end-effector task:

$$\mathbf{\tau }={\mathbf{J}}^{\top }\mathbf{F}+\mathbf{N}{\mathbf{\tau }}_0,\bar{\mathbf{J}}={\mathbf{M}}^{-1}{\mathbf{J}}^{\top }\mathbf{\Lambda }.$$

Notation flag. Khatib writes the operational-space inertia $\Lambda$ (joint-space inertia $A$); we use $\mathbf{\Lambda }=(\mathbf{J}{\mathbf{M}}^{-1}{\mathbf{J}}^{\top }{)}^{-1}$ per the notation.md “Additions” row and inertia $\mathbf{M}$ (not $A$). This $\mathbf{\Lambda }$ is the task inertia and is unrelated to the helix-waypoint set $\mathbf{\Lambda }=\{{\mathbf{\lambda }}_i\}$ in notation.md — a local glyph clash; do not conflate. $\bar{\mathbf{J}}$ is the dynamically-consistent inverse (notation.md), distinct from the mass-averaged ${\bar{\mathbf{J}}}_v$.

Source Support

• khatib1995inertial — primary; defines the macro-/mini serial decomposition (§5), proves the reduced-effective-inertia bound (Lemma 1, p. 27; Theorem 2 / Corollary 1, p. 28), gives the dextrous-dynamic-coordination null-space strategy (§5.3), and cites the free-flying-base application (Russakow & Khatib 1992, p. 29).

Related Topics

• coordinated_control — macro-/mini is the inertial-properties argument that justifies coordinating a slow macro base with a fast mini arm as one decoupled task-space controller.
• dynamic_coupling — the off-diagonal block ${\mathbf{M}}_{12}$ is exactly the macro–mini coupling inertia; the coordination strategy chooses null-space motion that does not let it disturb the end-effector task.
• redundancy_resolution — macro-/mini is solved as a redundant ($n>m$) system: the task is met in operational space and the spare DOFs (here, repositioning the mini stage) are assigned in the null space.
• operational_space_control — the underlying control framework; the macro-/mini result is stated entirely in operational-space (task-inertia) terms.
• augmented_object_model — the companion construct from the same paper, extending operational-space inertia from one serial chain to $N$ cooperating arms grasping an object.

Open Questions

• The construct assumes the mini stage is the lightweight one and minimizes instantaneous kinetic energy onto it; on our free-flying system the actuated base and arm may be comparably massed — does the macro/mini bandwidth split still hold, and how does it relate to the thesis circumcentroidal CoM/attitude+EE split rather than a proximal/distal split?
• The free-flying application is only cited (Russakow & Khatib 1992); base actuation costs fuel, so “carry out the task with the fast distal stage” may conflict with a fuel/effort criterion — what coordination objective replaces minimum kinetic energy when the macro stage is propellant-limited?
• The reduced-inertia bound is tight (equality) for prismatic mechanisms; what is the gap for a revolute arm on a flying base, and does it usefully bound achievable end-effector bandwidth?