The scariest moment in on-orbit servicing is contact. Two objects moving at orbital velocity, one trying to grab the other, with no second chance if the grab goes wrong. US10414053B2, "Robotic gripper for autonomous rendezvous and capture of satellites" (assigned to NASA), claims the mechanism that makes that contact survivable.

The claim is more specific than "a robot hand." Claim 1 targets a single, ubiquitous feature of launched spacecraft — the Marman ring, the clamp-band separation interface that joins a satellite to its launch adapter — and recites three jaws working as a set: "an outboard jaw which interfaces to an outer diameter side of the Marman ring," "an inboard jaw which interfaces to an inner diameter side," and "a palm which interfaces to a separation surface of the Marman ring." The grip is choreographed in two stages. First the jaws move "toward an opposing jaw in a direction parallel to the palm" and clamp the ring "across the outer diameter and the inner diameter." Then, "once a threshold clamp preload has been established, a second motion of the outboard jaw and the inboard jaw transitions to a downward motion, normal to the palm, which draws the Marman ring against the palm." Clamp first, then draw down to rigidize — the sequence is what turns a loose catch into a rigid, six-degree-of-freedom hold.

"A robotic gripper for rigidly grasping a section of a Marman ring of a satellite, the robotic gripper having an outboard jaw which interfaces to an outer diameter side of the Marman ring, a inboard jaw which interfaces to an inner diameter side of the Marman ring, and a palm which interfaces to a separation surface of the Marman ring."— U.S. Patent No. 10,414,053 source

The mechanism the claim protects is a gripper designed for capture under uncertainty. A servicing or debris-removal target often has no handle, no docking port, no cooperative markers — it was built to operate, not to be caught. But almost every satellite carries a Marman ring, and that is the handle this gripper exploits. The CPC is precise: B25J 15/0028 and B25J 15/0052 (robotic gripping mechanisms) joined to B64G 1/1078 (devices for use in spacecraft), with B64G 2004/005 marking the on-orbit-servicing context.

Here is the dependent-claim-is-the-moat point in concrete form. Anyone can claim "a gripper that catches a satellite." The asset is in the specifics of how it accommodates capture dynamics, and the dependent claims spell those out. Claim 3 constrains the inboard jaw to "only linear motion... normal to the palm," letting it follow the outboard jaw during draw-down but stay fixed during clamping. Claim 4 adds a "small draft angle" on the inboard jaw face that "encourages the inner diameter of the Marman ring to slide down" as the jaws close; claim 5 adds a "ramped tooth," and claim 6 notes the draft and tooth "approximate a V-guide" that coaxes a misaligned ring into position. Claim 8 gives the outboard jaw "a compound revolute geometry including a concave surface disposed between two convex surfaces" to make two points of contact, and claim 10 even calls out a "15° surface that matches a profile of a clamp band interface." Claim 11 describes an "underactuated mechanism" — rotary jaws on linear carriers that produce "simultaneous clamping and draw-down motion" from a single actuator — and claim 17 captures the payoff: correcting a ring "misaligned in three translational directions and in roll, pitch and yaw, simultaneously," through mechanical action alone. Claim 15 even enumerates specific launch-vehicle rings it fits, from Atlas V Type A to Delta IV. That is where prosecution narrows a broad idea into a defensible mechanism, and it is where this kind of grant earns its value.

Why NASA on the assignee line matters: this is foundational, government-developed IP in a field — on-orbit servicing, assembly, and manufacturing — that commercial players are now racing into. Government-origin capture mechanisms tend to become the reference art that the MacDonald Dettwiler and Northrop-class servicers build around or design past. The related MacDonald Dettwiler capture-mechanism grant US9764478B2 shows the commercial side of the same problem.

The two-stage motion at the center of claim 1 deserves emphasis because it is what separates this from a generic clamp. Clamping across the ring's inner and outer diameters first establishes a preload that centers and squares the ring; only then does the draw-down "normal to the palm" pull it flat against the palm to lock all six degrees of freedom. Claim 11's "underactuated mechanism" makes that two-phase choreography emerge from a single actuator — rotary jaws riding on linearly-driven carriers, where the clamping force itself (per claim 13) "induces a moment about a respective rotary jaw pivot axis which tends to roll the inboard jaw or the outboard jaw down." The draw-down is not separately commanded; it is a consequence of clamping. Fewer actuators means fewer ways for a one-shot capture to fail.

The self-aligning features are the other half of the moat, and they are aimed squarely at uncooperative capture. The inboard jaw's draft angle and ramped tooth (claims 4-5) form a V-guide that walks a misaligned inner flange into place; the outboard jaw's concave-between-convex face (claim 8) and cam-profiled lead-ins (claim 9) do the same on the outer diameter, even matching a 15-degree clamp-band surface (claim 10). The result claimed in claim 17 — correcting a ring "misaligned in three translational directions and in roll, pitch and yaw, simultaneously," through mechanical action — is precisely the property a servicer needs when it closes on a tumbling, handle-less target with no second attempt. The geometry absorbs the alignment error that the approach trajectory could not eliminate.

The honest limit: a granted gripper claim is a mechanism, not a demonstrated catch. Reduction to practice in orbit is a separate, expensive thing. But as a map marker, this grant tells you that the enabling step for the entire servicing economy — secure, self-aligning capture of an uncooperative target by its Marman ring — has been worked, claimed, and is now contested IP. Everything downstream of servicing, from life-extension to active debris removal, depends on solving exactly this grab.