The Biomechanical Inspiration for Soft Robotics

Nature has long been the primary source of inspiration for soft robotics. Biological organisms rely on compliant structures to navigate complex, unstructured environments — think of an octopus squeezing through a crevice, an elephant trunk wrapping around a branch, or a caterpillar undulating across uneven terrain. These movements are not based on rigid linkages and electric motors, but on soft tissues, tendons, and muscles working in harmony. Soft robotics aims to replicate this compliance, leading to machines that can safely interact with humans, handle delicate objects, and adapt to unpredictable surroundings. The two dominant bio-inspired actuation paradigms — tendon-driven systems and artificial muscles — each capture different aspects of biological motion, and their integration promises unprecedented mobility for soft robots.

Tendon-Driven Systems: Structure, Mechanics, and Control

Tendon-driven systems directly mimic the musculoskeletal architecture of vertebrates. A flexible, inextensible cable (the “tendon”) runs along a soft body segment and is tensioned by an actuator located remotely — typically a motor or a pneumatic cylinder. When the tendon is pulled, the body bends or contracts, producing motion with high precision and repeatability. The key advantage is that heavy actuators remain on the robot’s base or chassis, while the limbs stay soft and lightweight.

Key Components and Design Considerations

  • Tendons: Usually made of braided Dyneema, Kevlar, or high-strength polyethylene cable. Tendons must be low-stretch to transmit force efficiently, yet flexible enough to follow curved paths.
  • Actuators: Stepper or servo motors with winches, or linear actuators. The actuator’s speed, torque, and position feedback determine the fidelity of motion.
  • Pulleys and Sheaths: Reduce friction where tendons change direction. In soft robots, these can be embedded in the elastomeric body.
  • Backbone Structure: Some designs use a flexible central rod (e.g., nitinol wire) that the tendons wrap around, segmenting the body into discrete bending units.

Tendon-driven soft robots excel at tasks requiring precise positioning and force control. For example, the Harvard Soft Robotics Lab has developed multi-tendon continuum arms that can thread a needle or pour a cup of water. Their main limitation is that bending is restricted to planes defined by the tendon paths, and the system can suffer from cable friction and fatigue over many cycles.

Artificial Muscles: Materials that Move

Artificial muscles are active materials that change shape or size in response to an external stimulus — usually electricity, temperature, or pressure. Unlike rigid actuators, they resemble biological muscle in that they contract or expand along their axis, generating force while remaining compliant. Several families of artificial muscles have been developed, each with unique trade-offs.

Dielectric Elastomer Actuators (DEAs)

DEAs consist of an elastomer membrane sandwiched between compliant electrodes. When a high voltage is applied, electrostatic pressure compresses the membrane in thickness and expands it in area, producing large strains (up to 300%). DEAs can act as fast, silent artificial muscles, but they require high voltages (several kV) and thin films, which are prone to breakdown. Researchers at EMPA have used stacked DEAs to create soft robotic grippers that can lift objects ten times their own weight.

Shape Memory Alloys (SMAs)

SMAs like Nitinol (nickel-titanium) can be trained to “remember” a high-temperature shape. When heated above their transition temperature (e.g., via Joule heating), they contract strongly, pulling on the soft robot. SMAs offer very high force output relative to their size, and they are easily integrated as thin wires. The catch is slow cooling (and thus slow recovery), plus high energy consumption for heating. A soft robotic SMA-based octopus arm has been demonstrated by the MIT Soft and Self-Repairing Robotics Group.

Pneumatic Artificial Muscles (PAMs)

Also known as McKibben muscles, PAMs consist of an inner elastic tube wrapped with a helical fiber mesh. When pressurized, the tube expands radially and contracts axially, just like biological muscle flexing. PAMs are soft, safe, and can be made from cheap materials. They are widely used in rehabilitation exoskeletons (e.g., the Soft Robotics Toolkit by Harvard) and in biomimetic robots such as the soft robotic fish “SoFi”. The main drawbacks are the need for compressed air supply and the loss of efficiency due to friction and hysteresis.

Hydrogel and Hydraulically-Actuated Muscles

These materials swell or contract in response to changes in pH, temperature, or ionic concentration. Hydrogels are slower than DEAs or SMAs but can operate in wet environments, making them ideal for biomedical soft robots that must work inside the body. Hydraulically-driven soft muscles, using water or oil, offer high force and controllable stiffness, as demonstrated in the Octobot — a fully soft robot powered by a chemical reaction.

Integrating Tendon-Driven Systems and Artificial Muscles

Each actuation paradigm has strengths: tendons provide remote, high-precision force transmission, while artificial muscles offer distributed, compliant actuation with no moving parts. The most capable soft robots combine both in a hybrid architecture. For example, a robot might use a tendon-driven backbone to set the general posture (e.g., bending the spine) and then use artificial muscles — like DEAs or SMAs — to finer-tune the shape or to apply grip force at the end-effector.

Hybrid Structural Design

In one notable design from Cornell’s Organic Robotics Lab, a soft robotic arm uses multiple tendon cables for gross positioning and then deploys embedded PAMs in its “fingers” to gently wrap around objects. This combination allows the arm to handle both heavy items (by leveraging tendon tension) and fragile items (by using muscle compliance). Another approach uses shape-memory-alloy wires as active tendons: the SMA serves both as the tendon (transmitting force) and as the actuator (contracting upon heating), simplifying the overall system.

Control Challenges

Coordinating two different actuation modalities requires sophisticated control. Tendon-driven systems are typically modeled as continuum mechanics problems (e.g., constant-curvature kinematics), while artificial muscles introduce nonlinearities such as hysteresis, creep, and time-dependence. Researchers often use learning-based controllers or model-predictive control to blend the responses. For instance, in a soft robotic crawler, the tendon system might set the leg direction while artificial muscles control the foot’s compliance to avoid slipping.

Applications Driving Future Research

Minimally Invasive Surgery and Medical Devices

Soft robotic endoscopes benefit from both tendon-driven steering (for precise navigation through curvilinear anatomy) and artificial muscles for gentle tissue manipulation. Current prototypes can reach the colon or bronchial tubes with minimal trauma, and the integration of SMA-based “muscles” allows the tip to make fine adjustments during a biopsy. Rehabilitation exoskeletons increasingly use PAMs and tendon-driven cables to assist patients with stroke or spinal cord injury, providing natural joint torque with low inertia.

Agriculture and Food Processing

Harvesting delicate fruits like raspberries or avocados requires a gripper that can sense and conform to the fruit without bruising it. Hybrid soft grippers use a tendon-driven clamp to open the gripper wide, then artificial muscles (often DEAs) to apply a gentle, even pressure. Companies like Root AI (now AppHarvest) have tested soft robotic arms for such tasks, and the trend is toward more integrated tendon-muscle systems.

Search and Rescue in Confined Spaces

Search-and-rescue robots must squeeze through rubble, climb irregular slopes, and grasp victims or objects. Tendon-driven limbs can hold a payload, while artificial muscles embedded in the soles or pads provide traction and adaptability. The RoboClam is a classic example of a soft burrowing robot inspired by the razor clam, using a combination of tendon-like anchors and muscular contraction to move through soil.

Current Challenges and Research Directions

Power and Energy Density

Artificial muscles often have lower energy density than biological muscle. DEAs require bulky power supplies for high voltage; SMAs consume large amounts of electricity to stay hot; PAMs need compressors. Tendon-driven systems shift the weight to the base, but that limits autonomous operation. Researchers are exploring solid-state batteries, wireless power transfer, and chemical fuel cells that fit inside the soft body.

Durability and Self-Healing

Repeated bending and tension cycles cause fatigue in both tendons and artificial muscles. Tendons fray at the exit ports; DEA membranes can develop pinhole defects; SMAs can lose their memory over thousands of cycles. Self-healing polymers and redundant tendon paths are active areas of investigation. Self-healing artificial muscles, which re-bond microcracks when heated, could dramatically extend the lifespan of soft robots.

Manufacturing and Integration

Integrating tendons and artificial muscles into a single monolithic soft body is difficult with current fabrication methods. 3D printing of multi-material tendons, sensors, and muscles is one promising route. The ETH Zurich Soft Robotics Lab has developed a multi-step molding process that embeds SMA wires and tendon channels simultaneously, but scaling to mass production remains a challenge.

Conclusion: The Path Toward Lifelike Robotics

The convergence of tendon-driven systems and artificial muscles is moving soft robotics from a laboratory curiosity to a practical technology. By combining the positional accuracy of tendons with the compliant force generation of artificial muscles, engineers are creating machines that can walk, grasp, crawl, and even swim with an elegance approaching that of biological organisms. Future work will focus on enhancing power autonomy, improving control algorithms, and developing robust manufacturing methods. As these challenges are overcome, soft robots will increasingly assist humans in medicine, agriculture, disaster response, and exploration — performing tasks that were once considered possible only in science fiction.

The ultimate goal is a robot that moves as fluidly as a living creature, adapting instantly to its environment without rigid programming. Tendon-driven systems and artificial muscles are the two strands of the same biological thread, and together they are weaving the fabric of the next generation of lifelike machines.