Fin-Ray gripper that wraps around the object
Push a Fin-Ray finger against an object and it does the exact opposite of what you'd expect: instead of pulling away from the load, it curls toward the load and wraps around it. There are no sensors, no motors at each joint, not a single line of code adjusting the force. All the intelligence lives in the geometry: a triangle of flexible plastic that spreads pressure across its whole face and conforms to the shape of whatever it touches. It's one of those mechanisms where the part is the algorithm, and understanding why it works — and why it fails — comes down to understanding one thing: how a triangle moves when you push against one of its sides.
Why the finger closes instead of pulling away
The heart of the Fin-Ray effect is a triangle: two outer ribs that start from a wide base and meet at the tip, joined by a lattice of internal crossbars. The key is that this triangle isn't rigid. The ribs are thin and the crossbars act as a deformable lattice, so each vertex can rotate a little relative to its neighbors.
Imagine pushing on one of the two ribs at its midpoint, perpendicular to its face. In an ordinary beam, that push would simply dent the rib inward, and the tip would stay put or move away. Not here. The part is two ribs joined by a series of crossbars, and each crossbar is a strut of nearly fixed length: it barely stretches, it only pivots at its ends. When you push perpendicular to the contact rib, those crossbars can't accommodate it by stretching, so they force the two ribs to rotate relative to each other. That rotation accumulates crossbar by crossbar, from the load point toward the tip, and the result is that the triangle's vertex tilts toward the side under load. The finger curls over the object instead of backing away. The harder you push, the more it coils, and that progressive closing is exactly what tightens the grip, with no controller in the loop.
The shape of that motion is set by the geometry of the lattice, not by the material's properties: the path the tip follows depends on how the ribs and crossbars are arranged, not on how elastic the plastic is. Note one caveat, though: the joints here aren't free pins; they're the thin plastic regions themselves, which act as hinges by flexing. It's a pseudo-rigid-body mechanism: the geometry dictates the trajectory, but recovery, repeatability, and service life depend entirely on the material flexing and springing back. That's why the response is stable and repeatable only if you choose the material well, something the geometry alone won't guarantee.
Spreading the pressure keeps it gentle
A rigid finger touches an object at a point and concentrates all the force there; if the object is fragile, that point is where it breaks. The Fin-Ray does the opposite. Because the face that makes contact is a flexible rib supported by several crossbars, each crossbar absorbs its share of the load and the entire face conforms to the contour of whatever it grips. The result is pressure that's low, uniform, and spread across the whole flank of the finger, rather than a spike at one point.
That's what lets it pick up an egg and a wrench in the same configuration without reprogramming anything: the face curls until it wraps the irregular shape, and the force spreads across all the contact points at once. It's an adaptive grip achieved with geometry, not with control.
Crossbar density is the most direct design knob you have. Few crossbars leave long, lightly braced panels: the finger is softer, it coils sooner and with less force, ideal for delicate items. Many crossbars stiffen the lattice: the finger becomes firmer, grips harder, and deforms less against the same object. But you don't change only the overall stiffness — the spacing of the crossbars also moves where the curvature concentrates. More heavily braced spans bend less and push the wrapping toward the softer regions, so how many crossbars you use and where decides both the firmness of the finger and the point at which it starts to wrap. There's no correct number; there's a number for each task, and it pays to think about it in those terms — softness versus firmness — before tuning the rest of the dimensions.
Orienting it on the bed so it flexes instead of splitting
This is a monolithic compliant part: it isn't assembled, it's printed in one piece, and the joints are the thin plastic regions themselves acting as hinges. And like any FDM part that lives by bending, its Achilles' heel is anisotropy. The plastic is strong along the direction of the beads and weak between layers, where only the weld from one layer to the next holds it together; if you make a part flex in exactly the direction the layers are stacked, you don't bend it, you crack it open along the layers.
The rule here is blunt: print the finger lying flat, with its triangular plane parallel to the bed. That way the ribs and crossbars sit in the XY plane, their beads run in the direction each element is going to flex, and the deformation travels along the extrusion lines instead of pulling on the bond between layers. Print it on edge — with the layers stacked in the direction of the thickness that flexes — and every closing of the finger pulls directly on the interlayer weld: the crossbars tend to delaminate early, opening between two layers, and the part fails early. It's the same principle that governs any printed joint, as explained in Layer orientation for motion: the part is only as tough as the direction you stacked it allows.
Lying flat also gives you an advantage: the triangular voids of the lattice end up as in-plane cavities and don't call for support. What does call for it are the joints between rib and crossbar, because that's where the stress concentrates every time the finger closes. A sharp-cornered node is a stress concentrator, exactly the spot a crack ends up starting from; round those joints with a generous radius to spread the load, just as you round the root of a snap-fit arm.
With orientation settled, two dimensions refine the behavior. The rib thickness sets the stiffness of the walls that work in bending: thin ribs give a softer, more compliant finger that coils sooner, thick ribs a firmer one that pushes more before curling. And the triangle angle — how sharp the tip vertex is — moves the point along the finger where it starts to wrap: a more slender triangle wraps closer to the base and takes larger objects; a blunter one concentrates the curvature near the tip. Between the two, together with crossbar density, you decide where and how firmly the gripper closes.
Object size and how deep it should press in
The finger doesn't wrap around just any object: it wraps well around a band of sizes, and it pays to dimension for that band. If the object is too small relative to the finger, it barely indents the contact face, the lattice rotates little, and the grip is loose and barely wraps. If it's too large, the finger touches only at the tip, never gets to curl around, and works like a rigid stop — exactly what you wanted to avoid. The sweet spot is when the object fits loosely in the jaw opening but forces each finger to indent enough for the rotation to accumulate all the way to the tip.
That penetration — how far the object pushes into the finger's face relative to its rest position — is a first-order design parameter, as important as crossbar density or rib thickness. It's the initial interference of the grip: too little, and the finger doesn't coil; too much, and you force it past its useful range and overload the crossbars on every cycle. Dimension the jaw opening and the finger size so that the range of objects you expect to pick up falls inside that healthy press-in range, and you'll get a firm close without driving the lattice to its limit.
The material matters as much as the geometry
You can nail the perfect lattice and still end up with a useless gripper if you print it in the wrong material. A Fin-Ray lives by bending thin sections thousands of times, and that rules out rigid plastics from the start. PLA is the textbook example of what not to do: it's hard but brittle, doesn't tolerate repeated bending in a thin section, and the crossbars crack after just a few cycles. The combination of thin sections plus brittle material is the typical failure mode, and you don't fix it by thickening — thickening kills the softness that the whole mechanism depends on.
What you want is a material with real elasticity, one that withstands deforming and springing back many times without accumulating damage. TPU is the natural choice: it flexes, recovers, and handles cyclic fatigue well, exactly what a cyclically loaded living hinge calls for. Its downside is in sustained load — it's viscoelastic and tends to flow under constant stress — but for a grip that opens and closes it's an excellent compromise. Nylon (PA) is the other serious alternative: stiffer than TPU, with very good elongation at break and excellent fatigue life, it's in fact a reference material for living hinges when you want a firmer finger that recovers well. PETG ranks below: it has some toughness and performs far better than PLA, but in very thin sections with repeated flexing it eventually fatigues, so it's acceptable, not ideal. The rule is simple: in a part that flexes by design, the material's elasticity isn't a luxury, it's part of the mechanics.
And within TPU, Shore hardness is a design variable as real as any dimension. An 85A TPU is soft and coils with very little force — good for delicate items, weak for holding firmly; a 95A is considerably stiffer and prints with fewer headaches; a 98A or 63D is effectively semi-rigid. Picking the hardness locks in half the finger's behavior before touching the geometry, so decide it in parallel with the number of crossbars and the rib thickness, not afterward.
The failure modes and how to avoid them
Three ways to fail, and it pays to name them so you can design them out.
The first is fatigue of the internal crossbars. Every grip flexes those crossbars, and repeated flexing accumulates microdamage until one snaps. It's the failure mode that brittle material and overly thin sections suffer most: hence TPU or nylon, rounded joints, and dimensioning that doesn't drive any crossbar to its limit on every cycle. Here you aren't after maximum stiffness; you're after flex margin, just as a snap-fit arm is lengthened so it doesn't crack at the root.
The second is loss of shape from creep. If you leave the finger closed and loaded for a long time, the plastic slowly flows under constant stress and relaxes: the finger loses part of its shape and part of its closing capacity. TPU, being viscoelastic, is among the most prone to it. For a momentary grip it doesn't matter; for a gripper that has to stay clamped for hours, it does. Since TPU is exactly the material that flows the most, don't leave it pretensioned at rest; if you need to hold shape under sustained load, nylon will give you less drift. It's the same slow relaxation that loosens a press-fit over the months, described in Interference without cracking.
The third isn't about the part, but about the contact: the object slipping. The face that touches the object can let the load slide depending on the material and shape of what you're picking up; printed lying flat, that flank shows layer lines, a texture that helps somewhat but isn't always enough with smooth or heavy objects. The solution is to texture that contact face on purpose: add ribs, an engraving, or a rough finish to raise the friction and turn the geometric wrap into a grip that actually holds. It's cheap and resolves most slipping without touching the lattice.
What it's for and where to place it
The Fin-Ray is the finger you want when a gripper has to pick up varied objects without knowing in advance their shape or how much force to apply. Robot grippers that move from one part to another without reconfiguring, manual grippers, desktop manipulators that pick up whatever you put in front of them: in all of them the value is the same: an adaptive grip achieved with geometry instead of with programming. Where a rigid finger would force you to calibrate the force for each object, the Fin-Ray calibrates itself, curling until it spreads the pressure across the whole contact face.
That's why it's one of the emblematic mechanisms of compliant design for FDM, and at the same time a part that demands care in the three things we've already covered: correct orientation on the bed, a material with real elasticity, and a lattice dimensioned with margin. Get those three right and you'll have a gripper that wraps without sensors; get them wrong and you'll have a pretty triangle that unravels on the third squeeze.
The first step with any flexing finger is deciding how you stack it on the bed: that decision sets in advance whether the part bends or delaminates. For assembly clearances, see Tolerances for moving parts; for the golden rule of flexing, Layer orientation for motion.