Muscles are natural actuators that have spent millions of years perfecting the conversion of chemical energy into mechanical work. In this legacy, an entire field of biohybrid robotics is growing today: robots powered by living, lab-grown tissue, combined with artificial skeletons and precise mechanisms. Such systems have already shown they can crawl, walk, swim, or grasp objects, but in practice, they encounter two stubborn limitations. First, ranges of motion and generated forces are often modest compared to the needs of real-world tasks. Second, the interface between very soft muscle tissue and much stiffer skeletal parts is prone to mechanical damage and delamination, which limits durability and the possibility of repeatable operation. Precisely this "mechanical mismatch" was the inspiration for a team at the Massachusetts Institute of Technology (MIT) to introduce a proven strategy from biology into the polytechnic language: inserting artificial tendons between muscles and skeletons to transfer force more efficiently, making the assembly faster, stronger, and more durable.

A team led by an assistant professor in the Department of Mechanical Engineering, Ritu Raman, developed artificial tendons made of tough, adhesive hydrogels and connected them with a piece of lab-grown muscle tissue into a unique muscle-tendon unit. The ends of the artificial tendons are attached to the fingers of a small robotic gripper, while the central muscle serves as the drive. When the muscle is stimulated to contract, the tendons "pull" the fingers inward and convert the microscopic shortening of the muscle into a macroscopic, functional grip. Anyone working with actuators knows the trade-off between stroke and force – here, that trade-off shifts surprisingly favorably: compared to the same gripper where the muscle is directly attached to the skeleton, without tendons, closing is approximately three times faster, and the realized force about thirty times greater. Furthermore, the system showed an about eleven times better power-to-weight ratio compared to earlier prototypes driven by muscle without tendons and operated stably for more than 7,000 cycles without degradation of function – a figure that is already entering the territory of practical use.

Why the tendon changes the game

In natural biomechanics, tendons are not just "ropes" connecting muscle and bone. They are carefully tuned elastic elements that sit between soft and hard in terms of stiffness, reduce peak stresses at the interface, increase stroke, store and return elastic energy, and enable a more precise, less wasteful conversion of contraction into work. In contrast, in most earlier biohybrid robots, the muscle was stretched like a "rubber band" between two points on the skeleton. Such direct attachment wastes muscle tissue for mere attachment and often leads to tearing or detachment, especially when trying to extract greater force. Artificial tendons introduce what biology has already solved: a controlled series elastic interface that mitigates the mechanical difference and allows the muscle to work where it is most efficient.

In this MIT solution, tendons are shaped as thin, "cable-like" strips of high-toughness, strong-adhesion hydrogel. Their role is twofold. First, as elastic elements connected in series with the muscle, they increase the useful stroke of the gripper fingers for the same change in muscle length. Second, as an adhesive bond to the gripper skeleton, they distribute stresses over a larger surface area, avoiding critical stress concentration points that would otherwise tear the tissue or peel the joint. in actual operation, this means less "wasted" muscle and a greater possibility of precise, repeatable movement.

From spring theory to a gripper that actually works

Before material synthesis and assembly, researchers idealized the assembly as three springs connected in series: the muscle in the middle, tendons on both sides, and rigid gripper elements which can be represented in the model as springs of very high stiffness. The known stiffness of the muscle and structure served to analytically and numerically calculate the optimal tendon stiffness for the desired work – stiff enough to transmit force, but compliant enough to allow stroke. Based on these calculations, the hydrogel formulation and processing parameters were selected, and then the tendons were precisely cut into narrow strips that are easily guided over miniature "pulleys" on the gripper fingers. A grown piece of skeletal muscle was placed in the center using standard tissue techniques; the interfaces were designed so that the tendons chemically and mechanically "seat" onto both the living tissue and the synthetic skeleton.

When the muscle is stimulated (electrically, chemically, or optogenetically – depending on the design), the tendons transmit its shortening to the gripper fingers. The key is in adjusting the tendon pretension: a small initial tension removes slack and linearizes the system's initial response. In this configuration, the team measured about 3× higher closing speeds and approximately 30× greater forces compared to the version without tendons, and the gripper maintained such properties through >7,000 cycles without joint failure or loss of stroke. In parallel, the improvement in the power-to-weight ratio (~11×) was quantified, meaning less muscle tissue is needed for the same effect – crucial for miniature robots.

Modularity: a universal connector for different "skeletons"

Beyond the numbers, architecture is also important. Artificial tendons function as modules – interchangeable connectors between muscle actuators and various robotic skeletons. Once a set of parameters (length, stiffness, pretension, attachment method) is designed, the same module can be installed in different geometries: from micro-grippers for minimally invasive procedures, through agile grips for manipulating fragile specimens, to autonomous machines that adapt to unpredictable terrains. For development teams, this means faster iterations and better scalability – a new "muscle in the shape of a device" is not designed every time, but a standardized muscle-tendon unit is used, connecting to various skeletons like a "Lego brick".

Hydrogel that can stick, stretch, and survive cycles

For a tendon to be a credible engineering element, it must simultaneously be stretchable, strong, tough (i.e., resistant to crack propagation), and adhesive. This is a combination of properties that has long been contradictory in classic hydrogels. In recent years, formulations of tough hydrogels and composites have been discovered that reconcile this: polymer networks with energy dissipation mechanisms, with tunable crosslinking and adhesive functional groups. In this work, precisely such a hydrogel served as a "cable" that adheres equally well to the biological and engineering side of the interface. The result is a tendon that withstands thousands of cycles, evenly distributes stresses, and allows a larger part of the muscle to do the job for which it is evolutionarily optimized – generating force and stroke – instead of "acting" as glue.

Fine-tuning: stiffness and pretension

If the tendon is too stiff, the system stroke shortens and turns into a "hard" contact that again overloads the tissue interface. If it is too soft, force and energy are lost, and the gripper becomes slow and "spongy". Therefore, modeling was crucial: by selecting stiffness that balances these two extremes and adjusting pretension, an operating point is set that maximizes useful work per cycle. With the same philosophy, the team showed that force transfer from muscle to skeleton can be increased by about 29× when, in addition to the tendon, the stiffness of the skeleton itself is optimized. This confirms an intuitive, but often neglected truth from biomechanics: the actuator, elastic elements, and structure must be designed together for the system as a whole to be efficient.

Broader context: two parallel streams of progress

These results build upon two important streams of progress in the community. The first is the design of flexible, "spring-like" skeletons that increase the efficiency of muscle actuators by maximizing work per contraction through geometry and stiffness distribution. The second is the development of multi-directional artificial muscles – tissues that can contract in multiple directions (e.g., iris geometries), opening the way to more complex, "soft" movements. Artificial tendons are a logical connector between these streams: the muscle can produce sophisticated contraction patterns, the skeleton can elastically "respond" to them, and the tendons allow that dance to be mechanically sustainable and energetically useful.

Where we might see them

Medical microsystems. In minimally invasive surgery, a gentle, controlled grip, fatigue resistance, and exceptional miniaturization are required. Muscle-tendon units promise grippers and manipulators that work synchronously with tissue physiology, and are additionally potentially biocompatible.

Industry and laboratories. For manipulating fragile samples, cells, organoids, or soft goods, a combination of fine force and subtle stroke is suitable. Tendons allow the same muscle module to be "plugged" into different tools without redesigning the tissue.

Autonomous scouts. In inaccessible or risky environments, drives that are self-adaptive and durable are needed. Muscle tissue can be "trained", and interface damage is less when there is an elastic buffer – exactly what the artificial tendon offers.

Durability and reliability

Through fatigue testing, such assemblies maintained performance for more than 7,000 cycles. This is an important boundary: the transition from laboratory demonstration to a device that can be included in real processes begins only when mechanical reliability ceases to be a bottleneck. Hydrogel tendons have a dual contribution here – they retain adhesion during cycling and relieve the muscle of extreme peak stresses at the joint. This reduces the risk of tissue damage and extends the useful life of the entire assembly.

Materials science in the service of biomechanics

A broader look at the literature in recent years confirms that the hydrogel tendon is not an isolated invention, but part of a trend. Composites with anisotropic microstructures (e.g., with aramid nanofibers) have been developed that match, and even surpass, natural tendons in modulus, strength, and toughness. Such materials show that it is possible to combine high water content (biological compatibility), mechanical robustness (resistance to tearing and fatigue), and functional adhesion (connecting to living tissues and technical surfaces). MIT's contribution is that it does not treat such a tendon as a "replacement in medicine", but as an active mechanical element in robotics that can be mathematically designed and mass-produced.

From lab to practice: challenges that remain

For muscle-tendon units to find their way to application, several practical tasks remain to be solved. Standardization of muscle tissue culture protocols and connection with tendons is key for reproducibility between laboratories. Protective, "skin-like" sheaths are needed to keep the tissue from drying out and contamination, while simultaneously allowing gases and nutrients to pass. Control electronics must learn the "language" of biological actuators: controlling stimulation in a closed loop, compensating for property changes over time, and avoiding fatigue. For industry, cost per module will also be important: precisely why the modular approach – one muscle-tendon unit for multiple skeletons – is rational and economically attractive.

What follows

The next generation of biohybrid machines will likely combine multi-directional muscles (which can contract in multiple directions), elastic skeletons (which "amplify" work through geometry), and adaptive tendons (which shift performance to the desired regime by fine-tuning stiffness and pretension). With the help of additive manufacturing (3D printing of molds and "stamps" for guided fiber growth), such systems will be able to be rapidly manufactured and iterated. The "intelligence" of control will additionally come from learning from data: algorithms will optimize muscle stimulation patterns to achieve maximum efficiency with minimal fatigue.

Why it matters right now

The year is 2025, and soft robotics is increasingly emerging from laboratory demonstrations. At that moment, solutions combining strength, reliability, and modularity become key. Artificial tendons based on tough hydrogels show how a fundamental biomechanical lesson can be translated into a practical engineering advantage: instead of forcing soft actuators to "act" hard, an interface is built that is mechanically harmonized with both sides. The result is assemblies that grip faster, stronger, and longer from cycle to cycle – and with less muscle.

An important detail: a significant part of these findings was first published as an open preprint so that the community could quickly verify results and build upon methods, and studies were published in parallel confirming that the muscle itself can be shaped for multi-directional movements, and that skeletons can be designed as elastic "work amplifiers". The common denominator remains the same: instead of miniaturizing electric motors to the limits of physics, we learn from biology how to smartly connect the actuator, elastic element, and structure. Precisely for this reason – alongside muscles – tendons will increasingly be at the center of biohybrid robot design.