Engineers at Harvard have developed a 3D printing technique that produces soft robotic structures capable of twisting and bending in predictable patterns when inflated with air. The method, which pairs a rotating nozzle with two different materials extruded simultaneously, offers a new way to embed pneumatic channels directly inside printed filaments. What makes this approach distinct from earlier soft-robotics fabrication is the precision it brings to asymmetry: by controlling where internal air passages sit within each strand of material, the researchers can program exactly how a printed structure will deform under pressure.
How a Spinning Nozzle Builds Robot Muscles
The core innovation is a dual-material nozzle that rotates as it prints. A polyurethane shell forms the outer wall of each filament, while a hair-gel-like substance called poloxamer fills the interior as a temporary placeholder. Once printing is complete, the poloxamer is washed out, leaving behind hollow channels that act as pneumatic pathways. When air is pumped in, the channels expand, and because they are positioned off-center within the filament, the expansion produces bending or twisting rather than simple inflation. The technique is formally described in a recent preprint as rotational multi-material 3D printing of asymmetrical core-shell filaments, emphasizing how the spinning motion of the nozzle is central to shaping the internal architecture.
Three control parameters govern the final behavior of each printed piece: nozzle design, rotation rate, and print path. Adjusting the speed at which the nozzle spins changes the helical pitch of the internal channel, while the shape of the print path determines the overall geometry of the structure. Together, these variables let the team shape and orient internal channels with enough precision that the resulting devices bend and deform predictably when inflated, following motion profiles that can be simulated in advance. The result is something closer to a biological muscle fiber than a traditional pneumatic actuator, because the motion emerges from the material’s own architecture rather than from external joints or hinges.
The Team Behind the Technique
The work was led by graduate student Jackson K. Wilt and former postdoctoral researcher Natalie M. Larson, both working in the lab of Jennifer A. Lewis at Harvard’s School of Engineering and Applied Sciences. Lewis’s group has a track record in rotational multimaterial printing. An earlier iteration of the platform focused on helical filaments for dielectric elastomer actuator structures and structural lattices, establishing the foundational concept of using a spinning nozzle to create twisted internal geometries. The new work extends that concept from symmetric helical patterns to deliberately asymmetric ones, which is the key step that enables programmable bending and twisting in pneumatic soft robots.
The research has been posted as a preprint hosted by the arXiv platform, and a formal publication record with DOI 10.1002/adma.202510141 appears on the Lewis Lab publication page. The dual listing, one on an open preprint server and one through a peer-reviewed journal, suggests the team is moving quickly to make the findings accessible while the formal review process proceeds. For researchers outside Harvard who want to replicate or build on the method, the availability of a PDF through the lab’s own page lowers the barrier to entry considerably and complements the openly accessible version on arXiv.
Why Asymmetry Changes the Game for Soft Robots
Most pneumatic soft actuators rely on symmetric internal chambers. Inflate them and they expand uniformly, producing a simple ballooning motion. To get bending or twisting, engineers typically have to add external constraints, like stiffer layers on one side, or build complex multi-chamber designs that require separate air supplies for each degree of freedom. The Harvard approach sidesteps that complexity by baking the asymmetry directly into the filament during printing. A single air input can produce a compound motion, such as simultaneous bending and rotation, because the off-center channel geometry dictates the deformation path without the need for multiple valves or intricate tubing.
This matters for anyone building robots that need to interact gently with fragile objects or navigate confined spaces. Think of a surgical tool that must curl around tissue, or a search-and-rescue gripper that needs to conform to irregular debris. Traditional rigid actuators struggle in those scenarios because they move in discrete, hard-edged steps. A printed filament with embedded asymmetric pneumatics, by contrast, deforms smoothly and continuously, distributing forces more evenly along its length. The Lewis lab’s earlier work on helical filaments showed that rotational printing could produce artificial muscle-like structures optimized for contraction, but those symmetric designs were better suited to uniform motion along a single axis. The new asymmetric variant opens up a much wider range of motion profiles, from gentle hooks to corkscrew-like twists, using the same basic fabrication platform.
Limits and Open Questions
The method’s reliance on a fugitive core material introduces a post-processing step that could slow production. Every printed part must be soaked or flushed to remove the poloxamer before it can function. For a research lab producing one-off prototypes, that is a minor inconvenience. For a factory trying to produce thousands of soft robotic grippers per day, it could become a bottleneck unless the washing stage is automated and integrated into a continuous workflow. The technique also currently depends on polyurethane as the shell material, and it remains unclear from available sources whether other elastomers would behave the same way under the same printing conditions or whether they would require substantial re-optimization of ink rheology and curing parameters.
There are no published performance benchmarks, such as force output per unit of air pressure or cycle durability over thousands of inflations, in the materials summarized so far. Those numbers will be critical for anyone evaluating whether this approach can compete with established soft actuator designs from other leading robotics labs. Without head-to-head comparisons, the technique’s advantages remain largely qualitative rather than proven in application. The preprint details the fabrication method and demonstrates controlled deformation, but translating that into a reliable product will require additional engineering work on fatigue resistance, leak prevention, and integration with pumps and sensors that the current publications do not yet address.
What This Could Mean for Robotics and Prosthetics
If the durability and scalability questions can be answered, the implications extend well beyond laboratory demonstrations. In robotics, being able to print entire limbs or manipulators with embedded asymmetric channels could simplify design and assembly. Instead of routing external hoses or assembling multiple molded parts, engineers could print a single continuous structure that bends, twists, and stiffens in response to a small number of pneumatic inputs. Such actuators could be integrated into mobile robots that need to squeeze through tight gaps, underwater inspection systems that must conform to complex geometries, or warehouse grippers that handle everything from plastic bags to delicate produce without retooling.
Prosthetics and wearable devices are another promising frontier. Soft, muscle-like actuators that can be tuned to a user’s anatomy might one day offer more natural motion than rigid linkages and electric motors. A forearm sleeve with embedded asymmetric channels, for example, could assist with grasping or wrist rotation while remaining lightweight and compliant against the skin. Because the motion is programmed through geometry rather than through rigid joints, designers could create devices that feel less mechanical and more like an extension of the body. Realizing that vision will require careful attention to safety, reliability, and control strategies, but the underlying printing method gives researchers a powerful new tool for exploring these possibilities.
Open Access, Community Support, and Next Steps
The way this work is being shared also reflects broader trends in scientific communication. Hosting the manuscript on arXiv means that anyone with an internet connection can read the details without a subscription, and the platform itself is sustained in part by institutional members that contribute operating funds. Individual researchers and enthusiasts who find value in rapid access to preprints are likewise encouraged to support the service, helping to keep it free at the point of use. For early-stage technologies like rotational multi-material printing, that openness can accelerate feedback, replication, and creative adaptation in other labs.
For practitioners eager to experiment with the Harvard approach, the combination of the lab’s own publication page and arXiv’s documentation lowers practical barriers. New users can consult the platform’s help resources to understand how to access and download preprints efficiently, while the broader overview pages explain how the repository fits into the scholarly ecosystem. As more groups test the technique on different materials and applications, a clearer picture should emerge of where asymmetric core-shell filaments offer decisive advantages and where simpler actuators suffice. For now, the Harvard team has demonstrated a compelling proof of concept: by embedding carefully designed asymmetry at the filament scale, soft robots can be endowed with richer, more controllable motion, potentially reshaping how engineers think about artificial muscles in the years ahead.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
