Original paperSource
Reviewed by David Claude Bershadsky

Reviewed by David Claude Bershadsky, co-author · 2026-11-18

Make your paper interactiveCreate

Text from the interactive explanation

Digital Actuation Control of Soft Robotic Origami With Self-Folding Liquid Crystal Elastomer Hinges

David Claude Bershadsky, Tuo Zhao, Gláucio H. Paulino, Emily Davidson

Self-folding origami · Fold by wire

A flat circuit-board sheet that folds itself into a crane and flaps its wings when you tell it to.

Hinge energy Panel strain Body fold

Little energy stored: every hinge is close to the angle its temperature asks for.

A flat circuit-board sheet that folds itself into a crane and flaps its wings when you tell it to.

Simulated The crane's crease pattern, traced from the authors' OriCadLCE crane design: a four-point star with a small centre hole, slots at the neck and tail roots, and every mountain and valley fold as drawn there. Heater zones follow Fig. 1B and 6. A simulation of panels and hinges folds it. In thermal view, the panels cool with distance from the heated hinges, because heat is lost to the air as it spreads. Time sped up about 6×. Hinge targets follow the measured bend-versus-temperature curve of the Flex-PCB hinge (Fig. 2B), and the body stops just short of a complete collapse, as the real crane does. The energy plot comes from the same simplified simulation: hinge energy is high while heated hinges are still short of their angle, and panel strain peaks as the body passes through the middle of its fold.

01 · The muscle

A rubber that remembers order

A crease that folds itself needs a muscle: something that pulls, lets go, and pulls again. The authors use a liquid crystal elastomer (LCE), a soft, rubbery polymer network studded with stiff, rod-shaped molecules called mesogens.

As the ink is squeezed through a fine nozzle, it is sheared, and the rods line up along the direction the nozzle travels. Ultraviolet light then cross-links the ink and locks that order into the rubber. Drag across the print bed to lay down your own bead.

Predict: when you heat the printed strip past its transition temperature, will it get longer or shorter along the print lines?

Make a prediction, then slide the temperature up.

The schematic transition here is centred near 85 °C. For printed single layers, the paper measured an order parameter of 0.46 by X-ray scattering, near the ~0.45 typically seen as the maximum for printed LCEs, and a contraction ratio of about 0.42.

Stage 1, Ink: the ink is a liquid crystal oligomer, short chains built from rod-shaped C6M (RM82) mesogens joined by flexible thiol spacers. In the warm barrel the rods point every which way.

Blue rods are mesogens, cream curves are the flexible links between them, and amber bonds are UV cross-links. From the paper: the oligomer ink is made from the C6M (RM82) mesogen and the thiol EDDT plus small amounts of other reagents, reacted at 65 °C for 24 h. It is extruded at 40 °C through a 250 µm tapered nozzle, where shear and stretching flow align the mesogens along the flow, then cured under 365 nm UV light (about 35 mW/cm², roughly one hour per side). Sizes, numbers of molecules and speeds are schematic.

Heat scrambles the rods, and the rubber shortens along the print direction . Cool it, and both the order and the length come back. That is a reversible muscle. But a strip that only gets shorter is not a fold yet…

02 · The fold

Two layers make a hinge

To turn shrinking into folding, the authors print each hinge as two LCE layers with their filaments at right angles. A thin circuit-board sheet is sandwiched between them. One layer's filaments run across the crease; the other's run along it.

When the hinge heats, only the layer with filaments across the crease shortens across it. One face of the hinge gets shorter than the other.

Predict: when you heat it, which way will the outer panels swing?

The hinge bends toward the shortening layer. Swap the layer order and the fold reverses: valley or mountain . The paper sets each crease's direction simply by the order its layers are printed. Now, how do you heat a hinge?

03 · The old way

Why not just use an oven?

Until now, LCE origami was usually driven by its surroundings: an oven, a hot environment or a water bath. That works, but it heats every crease at once.

Here is the same crane, with its heaters unplugged and placed in an oven. One temperature dial controls everything.

Everything is below the switching temperature. Nothing moves yet.

Earlier work found clever workarounds. For example, it built folding sequences from two different LCEs that switch at different temperatures, then cycled the whole environment hot and cold. But global heating limits where folds happen and how fast. The authors note that environmental actuation is about an order of magnitude slower than the electrical approach in this paper.

One oven, one temperature, one shape. To choose which crease moves, each hinge needs its own heater. But a heater is a stiff sheet of metal and plastic, sitting exactly where the hinge needs to bend.

04 · The heater

A heater that doesn't get in the way

The paper's solution embeds a flexible printed circuit board (Flex-PCB) inside each printed hinge. It carries a copper heating trace: current through a resistor warms it, which is called Joule heating. Any sheet inside a hinge resists bending, so the authors tested cut-out shapes to find one that interferes least.

Each sample below is an LCE hinge with a different sheet inside, heated in an oven while it hangs from a clip. The last one has no heater at all and serves as the reference.

Predict: which heater-carrying design will bend the most at 100 °C?

Values traced by eye from the published plots (Fig. 2B, 2C); they are approximate averages, and the paper shows error bars. Torque was measured for three of the designs.

Hinge designs (top), printed samples (middle) and actuation at 100 °C (bottom). From Bershadsky et al. (2026), Fig. 2A, CC BY 4.0; cropped.

The serpentine cut-out leaves the hinge flexible, and a thinner 70 µm Flex-PCB keeps even more bend: about 113° at 100 °C, against about 150° for the heater-free reference. Its torque stayed similar to the reference under a ~1 g load. Its fibreglass (FR4) stiffeners are porous and bond tightly to the LCE, so this design did not peel apart at the edges.

05 · The isolation

Heat that stays where it's told

A heater in every hinge only helps if each hinge's heat stays put. On the Flex-PCB, every fold is surrounded by large areas of copper that the authors call dead zones or cooling pads. Copper spreads heat quickly and sheds it into the air, which keeps neighbouring hinges cool.

Power is set digitally by pulse-width modulation (PWM): the current switches fully on and off many times a second, and the fraction of time it is on (the duty cycle) sets the average power: P = i × V × duty cycle.

Let hinge A warm up, then switch the cooling pads off and compare hinge B.

The crane's Flex-PCB, with heater traces along every fold and labelled cooling pads (scale bar 20 mm). From Bershadsky et al. (2026), Fig. 1B, CC BY 4.0; cropped.

Thermal isolation means each fold can be addressed individually , and good heat dumping means each fold can relax quickly . The authors demonstrated a full cycle of 100 seconds, including extra time for even cooling. That is more than twice as fast as comparable electrically driven LCE actuators (200 to 300 seconds). But how do you know what temperature a hinge has actually reached?

06 · The loop

Closing the loop

Next to each heater sits a tiny temperature sensor (a thermistor). A microcontroller reads it many times a second and adjusts the duty cycle to reach a target temperature. This is closed-loop control.

The authors used a PID controller. It reacts to the present error ( P roportional), the accumulated past error ( I ntegral) and how fast the error is changing ( D erivative). Watch it climb a staircase of target temperatures. Then press You drive and see whether you can do better by hand.

The PID climbs from 25 °C to 100 °C in 5 °C steps.

The controller uses the paper's gains (KP 6.0, KI 0.2, KD 2.0), with output clipped from 0 to 255 (255 = 100% duty) and then capped. It drives a simplified thermal model of one 3.2 Ω hinge at 5 V. Playback is sped up. The measured staircases are in the paper's Fig. 5A.

In the paper's tests, steps of 5 °C from 25 to 100 °C held for 10 seconds left the hinge no time to settle, so it bent continuously. Held for 60 seconds , it settled at each step and climbed a staircase. The authors also capped the duty cycle at 20% or 30%. That slowed the response but let the motion follow the temperature more precisely.

A sensor plus a controller turns "heater on" into "go to 70 °C and stay there". That gives repeatable motion. Repeatable for how long, though?

07 · The endurance

Fifteen hundred folds later

The authors cycled one hinge between 30 °C and 100–120 °C under PID control for more than 1,500 cycles .

The swing range for cycles 0–500 is traced by eye from Fig. 5C (approximate). The paper plots range only for the first 500 cycles. For later cycles, it shows photographs.

The range dropped slightly over the first cycles, a "break-in" period, then levelled off. It jumped when the target was raised from 100 °C to 120 °C at cycle 100, then settled again. After the initial cycles, the authors saw no perceptible degradation . The test was stopped for lack of time, not because the hinge failed.

It is also frugal. Fifty cycles consumed about 1.3 watt-hours . The authors estimate that this hinge could cycle continuously for 2.85 hours on a single 1000 mAh lithium cell, which they describe as ideal for robots that carry their own power.

Reliable, individually addressable, digitally controlled folds. Now the fun part: what can you program with three of them?

08 · The program

Three heaters, many shapes

The authors' crane has three heating zones: body, left wing and right wing. Each combination, and each order of combinations, gives a different shape. Find all seven configurations the paper reports, then write a program of your own.

The crane starts as a flat four-point star. Its creases are traced from the paper's board and heater maps (Figs. 1B and 6). The heater map in the corner of the scene shows each hinge and turns red when its zone is hot. Spine, neck, tail, head and the four radial creases from the centre belong to the body zone. Two parallel hinges across the root of each wing belong to that wing's zone.

1 / 7 configurations found

Unactuated configuration all heaters off

Body heating body heaters on

Body folded body heaters on, after folding

Fully folded state all heaters on

Wing actuation wing heaters on

Left wing actuation left heaters on

Right wing actuation right heaters on

Now: Unactuated configuration (all heaters off)

Hinge energy: heated hinges not yet at their driven angle Panel strain: panels forced to flex Body fold

Switch on the body heaters and watch both curves as the crane folds.

Write a heating program

Each column is one step, about 16 simulated seconds long. Tick the heaters that should be on during that step.

Origami crane state-space diagram, with active-heater diagrams and thermal images (scale bars 3.0 cm). From Bershadsky et al. (2026), Fig. 6, CC BY 4.0; resized.

The Miura puzzle: three heaters, four shapes

The second robot is a square sheet patterned with a Miura-ori, a classic zigzag fold. Three straight creases cross four zigzag creases. The heater map (top left of the scene, traced from the paper's Fig. 7) shows two things layered on top of that pattern. Some creases form a diamond and are wired to two separate perimeter zones. Each zone owns a pair of opposite sides of the diamond. Zone A 's two sides lie on Miura creases, so it works in two shapes. Zone B belongs only to the box. The rest, the Miura creases, form the main zone. Where the diamond doesn't need a hinge, the sheet is simply split (dark lines). In the box, all four diamond sides fold the same way: the middle stays flat and the corners stand up. In the Miura, the zigzag needs two of zone A's crease segments to fold the other way, so those segments carry a hinge for each direction. The paper notes that a fold can be given both mountain and valley hinges. The paper reports four configurations: sheet , box (both perimeter zones), Miura (the main zone plus one perimeter zone) and a hybrid . Three of them follow from which heaters are on. The fourth depends on order. Can you find it?

Shape: Sheet configuration (all heaters off)

Try switching on both box zones.

Folded sheets like this one are often bistable: two stable shapes separated by an energy barrier. Heating tilts the landscape. The ball stays in its valley until that valley disappears, then it snaps across. Physicists call this a pop-through. Go from box to hybrid, then switch zone A on, and watch for the snap.

How these are computed: each crease is treated as a spring of equal stiffness. Its resting angle follows its heater, using the measured hinge curve (Fig. 2B). The energy is the average of ½(angle − resting angle)² over all 25 creases, from the simulated angles. Each panel shows the energy along a straight path from that shape (flag) to the sheet's current shape (ball). The ball is at the same height in every panel. It sits at a different distance from each flag. The brightest curve marks where the switched-on heaters lead: the shape with the lowest energy once its hinges reach full temperature. While the hinges are still warming, another well can be lower for a while. The hybrid is never where the heaters lead: it is held in place by the half-folded box creases, a lock this simple spring sum does not capture. The paper did not measure these energies.

Reconfigurable Miura sheet state-space diagram (scale bars 3.0 cm). From Bershadsky et al. (2026), Fig. 7, CC BY 4.0; resized.

Same three heaters, a different order , a different shape. The paper describes this as time- and thermal-dependent sequencing. With individually addressable folds, one sheet can store several crease patterns and switch among them on command.

09 · The design tool

Now design your own

The crane and the Miura sheet weren't wired up by hand, crease by crease. The authors built a design program, OriCadLCE , that turns a crease pattern into every file needed to fabricate the robot.

In OriCadLCE, you start by drawing the crease pattern and marking each fold as a mountain (red) or a valley (blue). Each crease is given a fold width . Where folds crowd together at a tight vertex, widening creases into bands would make them overlap, so the program automatically shortens each fold to avoid the clash. Then you add stiffener panels, adhesive tabs and electronics. From that one design, it generates the 3D printer paths for the LCE layers, with automatic pauses to insert the circuit sheet. It also exports vector files for the heater traces, stiffeners and outlines. The authors then draw the wires between heaters by hand in a circuit-board program and order the Flex-PCB.

Below is a much simpler version you can use right here. Draw creases between dots, choose mountain or valley and a heater channel for each, and pick a fold width. Then switch the heaters on and watch the same simulation used for the crane fold your design, following the measured hinge curve.

Design: 6 panels · 6 folds

Every vertex passes the flat-folding checks.

Simulation: Heating channel 1 (program)

"Simulate my design" heats channel 1, then 1 + 2, then all three, then lets everything cool. You can also switch channels on and off yourself. The small plot in the 3D view shows the stored hinge energy over time, coloured by channel. It is an illustrative model, not a measurement: each fold is a spring whose resting angle follows its heater, and the energy is the average of ½(angle − resting angle)². It spikes while hinges lag behind their heaters, and stays high when folds fight each other at a vertex that breaks the rules.

Mathematicians have worked out what any crease pattern needs in order to fold flat. Look at any point inside the sheet where creases meet. Maekawa's theorem: the numbers of mountain and valley folds there must differ by exactly two. Kawasaki's theorem: going around the point, every other angle must add up to 180°. The tool rings any point that breaks a rule. Folds around it fight each other, and the simulated sheet jams partway. These rules are general origami geometry, not results of this paper. The real OriCadLCE also has to stop the printer nozzle from hitting parts already embedded in the sheet. It does this by giving every element a height and thickness. That part is left out here.

Integrated design and fabrication workflow: (A) the OriCadLCE crease-pattern design and layer view, (B) generated LCE print paths, (C) generated heater and stiffener vectors, PCB design and fabricated Flex-PCB. From Bershadsky et al. (2026), Fig. 3, CC BY 4.0; resized.

A design program closes the loop. A new idea goes from a crease pattern to print paths and a circuit board without wiring each hinge by hand. That's what makes more complicated self-folding robots practical to build.

Same sheet, many shapes

A sheet covered in creases can become many different objects. The shape depends only on which creases fold, and in which direction. On a real sheet, that choice usually has to be made by hand, or built in when the sheet is made. Give every crease its own heater, as this paper does, and the choice becomes a switch. The sheet below is a generic pattern borrowed from a separate study: a square grid with an X in every 2 × 2 block. Tap a crease to make it a mountain or a valley, wire it to heater channel 1, 2 or 3, and then heat the channels.

No crease selected.

One wiring, three shapes.

Now: All heaters off

Generic sheet. The grid-with-X pattern is the "laminate sheet" in Fig. 1 of a separate study. That work folds sheets with magnetic fields and a heated plastic layer, not with LCE hinges. It is not part of Bershadsky et al. Only the crease geometry is reused here. The wiring and the three example shapes are ours. G. Unger, S. Shenoy, T. Li, N. Figueroa and C. Sung, "MORF: Magnetic Origami Reprogramming and Folding System for Repeatably Reconfigurable Structures with Fold Angle Control," accepted to the IEEE International Conference on Robotics and Automation (ICRA 2025); author copy, University of Pennsylvania repository.

Duck + boat. An OriCadLCE crease pattern from the authors' lab that did not make it into the paper. It is meant to fold into a duck with one set of creases and a boat with another. The creases were traced by eye from a screenshot of the design. Which creases belong to which shape, and the channel wiring, are our best guess. The simulated sheet only folds toward each shape.

What the simulation leaves out. Each hinge follows the measured Flex-PCB hinge curve (Fig. 2B), which tops out near 113°. That is short of a fully flat fold, so shapes stay partly open. Creases left off are unheated, passive joints: they stay roughly flat unless their neighbours force them to bend. Panels can pass through each other here, which a real sheet cannot do.

10 · What it brings closer

From one crane to a swarm

Back to the opening question. Every crease now has a muscle (a printed LCE hinge) and a wire (an embedded heater, sensor and digital channel). The authors look further ahead in two ways.

First, by moving the microcontroller, drivers and battery onto the Flex-PCB itself, they suggest, the robots could cut their tether. Second, by making self-folding origami cheap, fast and repeatable to fabricate, standardized self-folding "unit cells" could become building blocks for reconfigurable metamaterials , materials whose properties come from their folded structure, and for swarms of robotic cells with emergent behaviour.

One possible unit cell: a cube that grows

What might a unit cell look like? One next step being explored in the authors' lab, beyond what the paper reports, is a multistable expanding origami cube cut and folded from a single flat net. The net looks like a ladder: a column of squares, the spine, with an arm sticking out left and right at each of four squares. Red hinges fold as mountains and blue hinges as valleys. The yellow areas carry double-sided tape. First the spine rolls into a ring, and a yellow tab tapes it shut: the ring makes four of the cube's faces. Then the arms swing in without twisting. The top arm's end squares become the two remaining faces, and the other arms' yellow tips stay straight and tape onto the edges of those squares. That gives the expanded cube: six square faces joined by creased edge strips. Each edge strip has a valley crease across its middle, straight on some arms and diagonal on others, so the strips can fold like a Miura pleat. Twist the two end faces in opposite directions, like the ends of a candy wrapper, and the strips fold and tuck inward, pulling the whole cube in. At full contraction every strip is folded completely flat and the square faces meet edge to edge: a closed cube about a quarter as wide as the expanded one. The printed, Joule-heated hinges from this paper could give each crease its own muscle and wire.

Sketched, not simulated: panels are drawn rigid and the folds are placed by geometry, not computed from forces. Red bands are mountain folds, blue bands valley folds and yellow areas carry double-sided tape. The animation runs all the way between the fully expanded cube and the fully collapsed one, where each edge strip is folded flat inside and the square faces touch edge to edge. The real prototype below cannot yet reach that state, because its thick panels collide.

Link many cubes together and you get a soft robot built from voxels. In simulation, researchers have already evolved such voxel robots, made of contracting "muscle", soft "tissue" and stiff "bone" cells, into crawlers, walkers and jumpers (Cheney, MacCurdy, Clune & Lipson, GECCO 2013; Creative Machines Lab). Expanding cubes could be one way to build bodies like these.

Each cube expands, then contracts a little after the one behind it. Try setting the delay to 0°: does the inchworm still move?

The tube the team first set out to build

The team behind this work first set out to build a different structure: an origami tube whose cross-section can be re-programmed . It comes from a 2016 paper by Evgueni Filipov, Glaucio Paulino (also an author of this paper) and Tomohiro Tachi. Their tube folds flat along its length like a bellows, and opens again, while every panel stays a rigid, flat parallelogram. A few strips act as switches . When the tube is fully open, these strips lie flat, and each one can then fold down one of two ways, like a binary bit: 1 (sloping up) or 0 (sloping down).

One switch on the top and one on the bottom gives two shapes. Split the switch strip into more pieces and the count explodes. The tube only closes if the top and bottom rows have the same number of 1s, and the paper shows that this leaves (2n)!/(n!)² configurations for n switches per side: 2, 6, 20, then 70 for four. Their four-switch tube can take 70 distinct cross-sections. Try predicting before you count: with two switches per side, how many settings of the four bits are allowed?

Such a tube has more than a hundred creases, and each must fold the right way at the right moment. That is exactly what this paper's hinges offer: each crease could be a printed LCE hinge with its own heater channel. The tube paper itself says its switches will need "new methods for rapid or remote actuation". The demo below is our illustration of one of their tubes. It shows a direction for future work, not a result of the main paper.

The flat pattern. The tube cannot be folded from one sheet, so it is cut as one strip per side and taped along the long edges, as the authors did for their paper models. Red solid lines are mountain folds, blue dashed lines are valley folds, seen from outside. Flipping a switch reverses the folds on its pale strip and its edges.

Tap a switch to flip it between 1 (slopes up) and 0 (slopes down). The tube can only take the new shape if the upper and lower rows have the same number of 1s.

Pressed flat along its length: zero volume, yet every panel is still a flat, undistorted parallelogram.

Geometry computed from the paper's construction: the cross-section of its Fig. 2a (side slopes 30°, 90° and 125°), copied along a zig-zag at 60° for ten segments. For two to four switches, the flat 90° side is split into equal switches, 0.5 long each, an arrangement we chose; we checked that all of their configurations give a cross-section that does not cross itself. The counting rule is the paper's equation 4.2. Folding keeps every panel rigid, which leaves a single motion. Volume and area are measured on that geometry. The tube is drawn with zero thickness, like the paper's own models. Source: E. T. Filipov, G. H. Paulino & T. Tachi, "Origami tubes with reconfigurable polygonal cross-sections", Proc. R. Soc. A 472: 20150607 (2016), PDF.

11 · The evidence

What the paper actually showed

How one is made

The fabrication method, layer-by-layer direct ink writing , merges two industries: 3D printing and ordinary circuit-board manufacturing.

Design. A crease pattern is drawn in the authors' software, OriCadLCE. It automatically generates printer tool paths for every mountain and valley fold and vector files for the heaters and stiffeners, while avoiding collisions with the embedded sheet.

Print the first LCE layer. The ink is extruded at 40 °C through a 250 µm nozzle, which aligns the rods along each filament.

Pause and insert. The printer stops, and a commercially made Flex-PCB with heaters, thermistors and FR4 stiffeners drops onto alignment pegs.

Print the second layer on top, with filaments at right angles, sandwiching the circuit board.

Cure. 365 nm ultraviolet light cross-links the network, for about an hour per side.

The contribution is not one robot. It is a repeatable recipe: hinges that fold reliably, a heater in each crease, and design software that turns a crease pattern into fabrication files. Together, that brings programmable, self-folding machines, and perhaps whole swarms of them, a step closer.

Sources & credits

David C. Bershadsky, Tuo Zhao, Glaucio H. Paulino and Emily C. Davidson. “Digital Actuation Control of Soft Robotic Origami With Self-Folding Liquid Crystal Elastomer Hinges.” Advanced Functional Materials 36 (2026): e25150. doi.org/10.1002/adfm.202525150

The article is open access under a Creative Commons Attribution 4.0 licence. Photographs and thermal images are reproduced from its Figures 1B, 1C, 2A, 3B, 4C, 5D, 6 and 7, cropped or resized. Curves in the hinge lab and the endurance chart were traced by eye from Figures 2B, 2C and 5C and are approximate.

The crane, oven, Miura and PID scenes are simplified simulations. The crane uses the crease pattern and mountain/valley assignments of the authors' OriCadLCE crane design (also shown in Fig. 3A), with heater circuits from Fig. 1B and 6. Its thermal view spreads heat from each hinge with a simple exponential fall-off, and its energy plot comes from the same simplified simulation. The Miura sheet uses the pattern and heater circuits traced from Fig. 7. A simulation of rigid panels and hinges computes their folded shapes, and the hinge angles follow the measured Fig. 2B curve. The Miura state rules follow the paper's description of Fig. 7. The mini design tool is a simplified version of OriCadLCE: it folds your design with the same simulation and measured hinge curve, but its print-layer and circuit previews are schematic. The Miura energy landscape is an illustrative cartoon of bistability, not a measurement. The printing, bilayer, heat-flow and expanding-cube scenes are illustrations of the mechanisms and rules described in the paper, not reconstructions of the authors' data. The authors' datasets, raw videos and software are deposited at Princeton Data Commons, doi.org/10.34770/q55n-0m58 (CC BY-SA 4.0).

Further reading for the future-work section, separate from this paper: E. T. Filipov, G. H. Paulino & T. Tachi, "Origami tubes with reconfigurable polygonal cross-sections", Proc. R. Soc. A 472: 20150607 (2016), doi.org/10.1098/rspa.2015.0607. D. C. Bershadsky & G. H. Paulino, thick flat-foldable twisted-tube origami, engrXiv (2024), doi.org/10.31224/3811. The expanding cube, its net and the prototype video come from follow-up work in the authors' lab and are not results of this paper. The LCE printing diagram is an illustration of the process described in Sections 2.1, 4.2 and 4.3.

Topics in this iPaper

Engineering
Highly repeatable, closed loop, digital actuation control is achieved through integrated Joule heating of liquid crystal elastomer hinges.

Abstract, page 1

Chemistry
The liquid crystalline oligomer ink is based on the C6M

Section 4.3 LCE Synthesis, page 10

soft robotics
Soft robotics has sought to integrate seamless, programmable shape morphing into dynamic, complex structures.

Abstract, page 1

liquid crystal elastomers
reversible shape memory materials such as liquid crystal elastomers (LCEs) have enabled self-folding origami to achieve high work output

Section 1 Introduction, page 2

direct ink writing
Layer-by-layer direct ink writing (LBL-DIW) allows replicable fabrication of origami with integrated programmable actuation control.

Section 2.1, page 2

origami
Origami offers a vast library of reconfigurable 3D structures that can be folded from sheet materials.

Section 1 Introduction, page 1

joule heating
Pre-fabricated Flex-PCBs allow targeted Joule heating of individual hinges through integrated circuit design

Section 2.1, page 2

closed loop control
PID feedback control was characterized on a single hinge to understand the mechanical response of the system to the control loop.

Section 2.3, page 7

Digital Actuation Control of Soft Robotic Origami With Self-Folding Liquid Crystal Elastomer Hinges — Interactive explanation | iPaper.ai