Fullscreen HTML5Med Res (??MB)HTML4

SOFT ROBOTS by Dr David Maddison, VK3DSM Prof. Cecilia Laschi’s robotic octopus from the National University of Singapore. Source: Jennie Hills, The Science Museum, London Imagine a robot navigating through a disaster zone, squeezing through rubble like a worm to find trapped survivors, or a medical soft gripper handling human tissues or even squeezing through veins and arteries. This is beyond the capabilities of traditional robots, but within the emerging field of soft robotics. S oft robotics is still mostly confined to laboratory research, but there are emerging areas of commercial devices. Soft robotics is a blend of biology, materials science, engineering and AI to create flexible, adaptable machines that can mimic living organisms. Unlike conventional robots with fixed joints and stiff links, soft robots are built primarily from compliant, deformable materials like elastomers or elastomer-like materials (eg, silicones and hydrogels) that can stretch, bend, twist and squash. This flexibility enables safer interaction with humans, adaptability to unpredictable environments and the imitation of natural movement, from 12 Silicon Chip the muscular hydrostats of an octopus tentacle to the crawling of a caterpillar. Soft robotics might be used for just part of a traditional robot, such as a gripper at the end of a conventional robot arm (an ‘end effector’), to pick up delicate items like fruits and vegetables, or perhaps for legs, as with the robot turtle we will discuss. The emergence of soft robotics is driven partly by the demand for new and more versatile robots for applications not suitable for existing robots, such as delicate surgery; operating in restricted, unstructured environments like disaster zones; or crowded environments like factories. Soft robotics is being made possible with technologies like: Australia's electronics magazine • advanced 3D printing techniques that allow elastomers and other specialised materials to be printed • artificial intelligence and microfluidic ‘brains’ for control • novel materials, such as those with self-healing and stimuli-­ responsive properties • diverse actuation mechanisms (pneumatic, hydraulic, electrostatic and more) to provide powerful yet gentle motion • flexible (bendable) sensors and electronics This article explores the materials, mechanisms and biological inspirations behind soft robots, plus examines their growing applications across medicine, manufacturing, exploration siliconchip.com.au and more. We will look at the role of microfluidics in some designs, and investigate real-world commercial products, before addressing remaining challenges and glimpsing at possible future applications. Fluidics is a concept that will come up throughout this article. We published a detailed article on its principles in August 2019 (siliconchip.au/ Article/11762). Characteristics of soft robots Soft robots are constructed primarily from soft, highly deformable and compliant materials that enable them to bend, twist, stretch, and conform to complex shapes. This inherent softness and conformability give their movements a fluidity far more reminiscent of biological organisms than traditional rigid robots. Rather than relying solely on conventional electric motors and geared joints, as with rigid robots, soft robots frequently employ pneumatic or hydraulic actuators (inflating or pressurising fluid-filled chambers) to generate motion. Control can come from traditional embedded electronics and/ or AI, or in some designs, from microfluidic logic circuits for certain tasks like locomotion. While microfluidic logic circuits can provide basic rhythmic or sequential locomotion capabilities (eg, alternating leg motion like an insect or simple oscillating gaits), they cannot perform the complex, high-speed processing or adaptive decision-making possible with traditional CPUs, GPUs, NPUs and AI algorithms. Historical evolution and biological inspiration The first entirely soft and autonomous robot is generally regarded as the Octobot (2016), which will be discussed later. It is made of soft silicone gel and uses pneumatic actuation to control its limbs from a microfluidic logic circuit. As implied by its name, it is inspired by the octopus. Prior to the Octobot, there were various foundational developments, including smart and flexible materials, advanced fabrication techniques like 3D printing of soft materials, and new methods of mathematical modelling and control for compliant systems. Pneumatic actuators, as used in modern soft robots, trace their origins back to 1957 with the McKibben siliconchip.com.au artificial muscle (also known as the pneumatic artificial muscle or “air muscle”). These compliant actuators featured a rubber bladder inside a braided sleeve that contracted when pressurised, mimicking biological muscle contraction. That was one of the earliest uses of soft, deformable materials for actuation. The artificial muscle was initially developed for artificial limbs and orthotic devices to help paralysed patients grasp objects. Although industrial robots existed at the time (eg, Unimate, described in our May 2017 article; siliconchip.au/ Article/10641), McKibben’s innovation laid the foundation for the compliant pneumatic actuators that power many present-day soft robots. In the early 1990s, researchers including S. Shimachi and M. Matsumoto pioneered the use of silicone micro-actuators and soft compliant fingers for robotic manipulation. Their work modelled deformation, friction and grasping stability of deformable silicone fingertips. This marked one of the earliest systematic investigations into compliant, soft end-effectors, demonstrating improved adaptability, reduced object damage and enhanced force control compared to rigid fingers. This research paved the way for the modern soft robotics era, influencing later pneumatic soft grippers and compliant actuators. During the 1990s, robotics research also saw growing interest in bio-­ inspired designs, with projects such as Joseph Ayers’ lobster-like robots, which drew on animal locomotion for inspiration. However, these early systems remained rigid, using hard exoskeletons and conventional motors rather than compliant materials. Together, pneumatic artificial muscles, the systematic modelling of soft silicone actuators and the increasing emphasis on biological inspiration laid the conceptual and technical groundwork that would enable the emergence of modern soft robotics in the following decade. Materials Many different materials are used in the fabrication of soft robots. Elastomers and silicones are used for their high levels of flexibility, stretchability and formability. Shape memory materials are also used, which change shape Australia's electronics magazine in response to heat, light and electric fields. They may be polymers (plastics) or metals. Electroactive polymers deform when an electric field is applied. Hydrogels can be engineered with many desired properties, such as responsiveness to environmental parameters like pH, humidity, light and magnetic fields. A hydrogel is a polymer material that can retain large amounts of water, giving them a soft, flexible consistency similar to living tissue or jelly. Elastomers such as silicone (eg, PDMS & Ecoflex) and thermoplastic polyurethane (TPU) are commonly used in soft robotics, as are hydrogels. Shape memory polymers and metal alloys are ‘smart’ materials that can be deformed into a temporary shape and then, by the action of some stimulus, return to their original shape. For polymers, the stimulus may be heat, light or electricity; for alloys, it can be heat. Piezoelectric polymers can convert motion to electricity and vice versa. Self-healing materials are being researched for soft robotics, to enable robots to autonomously repair damage and extend their operational lifespan in challenging environments. These materials, typically based on polymers like silicones or hydrogels, incorporate reversible chemical bonds, embedded microcapsules or vascular networks containing healing agents. They activate upon crack formation and release monomers to polymerise or flow to seal breaches, a bit like self-sealing tanks on military aircraft or the platelets in our blood. When damage occurs, such as cuts, punctures or fatigue, the material can restore its mechanical properties. In soft robotics, self-healing enhances their durability for applications like medical devices (eg, catheters that can repair themselves inside the body), disaster-response robots operating in harsh conditions, or wearable exosuits subject to wear. Fabrication Soft robotic components are typically either moulded or 3D printed. Mould casting is a technique where liquid elastomers are poured into moulds, then cured and removed. Channels or chambers can be cast within the part for pneumatic or hydraulic actuation. Multiple parts July 2026  13 Fig.1: two flexible parts can be moulded, then dipped in adhesive, glued together and cured. Fig.2: a dielectric elastomer actuator changes its dimensions in response to an electric field. Original source: www.digikey.com/en/maker/projects/ diy-soft-robotics-dielectric-elastomerdot-actuator/5b77674365634d86b1f97 87fa4501c9b Fig.3: a dielectric elastomer actuator configured so the tip bends as the electric field is cycled. Source: www. digikey.com/en/maker/projects/diysoft-robotics-dielectric-elastomer-dotactuator/5b77674365634d86b1f9787f a4501c9b can be adhered together – see Fig.1. This technique was used for the Octobot and is commonly used today for robot grippers and many other components. Mould casting is used for PneuNets (pneumatic networks), a very common type of soft robotic pneumatic actuator, described later. 3D printing includes a variety of techniques such as FDM/FFF (fused deposition modelling and fused filament fabrication) and DIW (direct ink writing) to extrude materials like TPU, silicones and hydrogels. Vat photopolymerisation (including SLA [stereolithography] and DLP [digital light processing]) is also used to cure liquid resins with light. Material jetting can be used to deposit multiple materials simultaneously, including soft and rigid materials, possibly even embedding electronics. “4D printing” is an emerging technique in which a material changes shape after fabrication. Such a device could be fabricated using the moulding or 3D printing techniques described above. Shape change is brought about under the influence of heat, light, moisture etc. This technique enables self-folding or self-­ assembling robots, often biomedical or miniature types. Microfluidic controls for soft robots are fabricated using a variety of techniques. These include subtractive manufacturing (where materials are removed to create microchannels), moulding, micromachining and 3D printing. Actuation mechanisms As typical motors are usually incompatible with soft robots, actuators to generate movement typically rely on methods involving electrical or fluidic activation and sometimes magnetic activation. Electroactive polymer actuators use various combinations of electrodes, insulating polymers, conducting polymers and piezoelectric polymers to achieve movement. Electroactive polymer actuators include: Dielectric elastomer actuators consist of a thin elastomer film sandwiched between two compliant (stretchable) electrodes. When a high voltage is applied across the electrodes, the resulting electric field compresses the elastomer and causes it to expand in area, producing shape change and mechanical work (see Fig.2). The elastomer film is typically tens to hundreds of micrometres (µm) thick, and operating voltages range from tens of volts to several kilovolts. In practical devices, multiple layers are stacked to achieve greater force and stroke. A common configuration uses two elastomer layers with a shared ground electrode in the middle and separate high-voltage electrodes on the outer sides, as in Fig.3. This arrangement mimics antagonistic muscle pairs: applying a voltage to one layer causes it to expand (contracting the opposing layer), enabling bidirectional bending or linear motion. Examples of dielectric elastomer grippers are shown in Fig.4. An article on how to make your own device can be found at siliconchip.au/ link/acao Liquid crystal elastomer actuators are advanced stimuli-responsive materials used in soft robotics for their ability to undergo large, reversible shape changes, often changing dimension by 50% or more when triggered by external stimuli such as heat, light or electric fields. These materials combine the ordered molecular alignment of liquid crystals with the elasticity of polymer networks, allowing programmed molecular orientation during fabrication Ground Vchuck Vactuator P Electrode terminals DEA units Vchuck = 0 Fig.4: examples of dielectric elastomer grippers and structure. Source: www.mdpi.com/2076-3417/10/2/640 14 Silicon Chip Australia's electronics magazine siliconchip.com.au (eg, via 3D printing or alignment techniques) to dictate precise deformation patterns like bending, twisting, or contracting. Challenges remain in response speed, force output and durability. Fig.5 shows a variety of liquid crystal polymers and the shape transition of liquid crystal elastomers. The liquid crystal main chain polymers (LCP) are shown, then a liquid crystal polymer network (LCN), then a liquid crystal elastomer (LCE). Only LCEs can perform a shape change. The difference between LCNs and LCEs is that LCNs have many more cross-links between the polymer chains (too many to allow a shape change). Ionic polymer actuators bend or deform when a voltage is applied, mimicking muscles by moving ions within a polymer membrane, causing swelling in one area and shrinkage in another, resulting in motion. Key types are ionic polymer metal composites and ionic polymer gels. They work through the application of a low voltage, causing ions to migrate to the oppositely charged electrode, as shown in Fig.6. Piezoelectric polymers are just like ceramic piezoelectric crystals, such as quartz. Motion is converted into electrical energy or vice versa. The main difference is that a polymer is used rather than a ceramic. PVDF (polyvinylidene fluoride) is a common piezoelectric polymer. Not only can such polymers be used for actuators, they can be used for sensors as well. A single sheet of polymer will not generate enough motion, so typically they are assembled in two layers. When an electric field is applied, one layer shrinks and the other expands, causing motion, as shown in Fig.7. Conducting polymers (see our November 2015 article on the topic) are polymers such as polypyrrole that are intrinsically electrically conducting and don’t rely on metal or carbon fillers to render them conductive. Ions can be moved into and out of them in an appropriate solution, causing them to change shape, as in Fig.8. Hydraulic actuators use water or oil to inflate a bladder or similar structure. Magnetic actuators generate motion via an external magnetic field interacting with magnetic materials inside the robot. Photoresponsive actuators react to light by changing shape, stiffness siliconchip.com.au Liquid crystal mainchain polymers (LCPs) (A) Liquid crystal polymer networks (LCNs) Liquid crystal phase (B) Liquid crystal elastomers (LCEs) Isotropic phase Cooling Heating Fig.5: liquid crystal polymers. LCEs can change shape upon heating, cooling or some other stimulus. Source: https://encyclopedia.pub/entry/history/ show/60582 Fig.6: the operation of ionic polymer actuators. Original Source: www.mdpi.com/20734360/17/6/746#polymers-17-00746-f002 Fig.7: an activated PVDF bimorph showing motion from the vertical position. Source: https://physics. montana.edu/eam/polymers/ bimorphs.html Fig.8: a conducting polymer (polypyrrole) actuator with motion as the voltage is switched. Australia's electronics magazine July 2026  15 or volume. These include materials such as: Liquid-crystal elastomers, which can change length by up to 50% when exposed to light (mentioned above) Polymers containing photosensitive compounds that bend or twist on light exposure Photothermal composites, which contain layers of materials with different properties in which the structure bends when exposed to light Hydrogels with photosensitive compounds that change stiffness when exposed to light Pneumatic actuators are probably the most common type of actuators for soft robots. They use compressed air, another gas or the decomposition of a fuel like hydrogen peroxide to generate gas. A PneuNet is an example of such an actuator. It typically has two layers or areas; one that is extensible with an internal air chamber, and another inextensible layer or area. When the extensible chamber is inflated, the assembly bends, constrained by the inextensible layer or area – see Fig.9. They can be fabricated by casting in 3D-printed moulds. Thermally responsive actuators use shape memory alloys, polymers or thermally responsive hydrogels that change shape in response to heat. Sensors Sensors for soft robots are chosen for their ability to maintain compliance and stretchability so they can be integrated into soft, deformable bodies. Examples include: Stretch/strain sensors are thin, stretchable films or fibres that change electrical resistance or capacitance when deformed. Materials include carbon black in elastomers, or lowmelting-point liquid metal alloys like EGaIn (liquid eutectic gallium-­ indium) in microchannels. Applications may include force estimation with grippers and soft exosuit strain monitoring. Soft pressure/tactile sensors measure contact pressure or distributed force via changes in resistance, capacitance or optical properties. Techniques include piezoresistive (conductive foam/rubber), capacitive (elastomer layers with flexible electrodes) or optical (light intensity change through deformable waveguides). Applications include gentle grasping of fragile objects (eg, fruit, eggs), human-robot safe interaction and texture discrimination. A commercial example is the grippers from Soft Robotics Inc, which have embedded soft tactile sensors for slip detection. Embedded optical fibre sensors detect strain, curvature or temperature by changes in light wavelength or intensity. Applications include surgical catheters or soft robot arms/ tentacles, and many other snake-like soft manipulators for minimally invasive surgery. Soft magnetic sensors use Hall-­effect Fig.9: a variety of configurations of PneuNet type actuators to give different shapes or motions. Source: https://elveflow. com/microfluidicreviews/soft-robot/ sensors or magnetometers to detect changes in magnetic fields from embedded soft magnets or ferrofluids. They can be used for curvature/angle sensing in pneumatic actuators or soft exosuits for joint angle measurement. Ionic/electroactive polymer sensors use ionic polymer-metal composites (IPMC) or dielectric elastomers to generate a voltage/current when bent or stretched (self-sensing). Applications include self-sensing actuators (one material acts as both the actuator and sensor). Emerging/bio-inspired sensors such as hydrogel-based chemo-­sensors to detect pH, temperature or specific chemicals (eg, for environmental monitoring); bio-hybrid sensors such as living cells (eg, muscle cells) integrated with soft robots for chemical or biological sensing; stretchable cameras (miniature soft cameras for visual feedback, eg, in medical or underwater soft robots). These sensors are often integrated directly into the soft elastomer body during fabrication (3D printing, moulding or embedding), making soft robots sensor-filled from the inside out. Mathematical modelling Because of the ability to continually deform with infinite degrees of freedom, soft robots need different control and modelling strategies compared to traditional rigid-body robots. New mathematical models have been developed based on “Cosserat rod theory” to predict the complex non-linear behaviour of such robots (see Fig.10). Traditional and finite element analysis are also used. Cosserot rod theory is a mathematical framework that applies to soft, deformable slender structures like rubbery tubes to accurately model behaviours like bending, extension, shear and twisting. Normal structural models cannot account for these Fig.10: a variety of soft robot elements that are modelled with Cosserat rod theory. They would be difficult or impossible to model using other methods. Source: www. researchgate.net/publication/383153694 16 Silicon Chip Australia's electronics magazine siliconchip.com.au Soft controller Fuel reservoirs Reaction chambers Actuators Vent orifices Fuel inlets Upstream check valves Pinch Downstream valves check valves Outlets Figs.12 & 13: Octobot, the first fully soft, electronics-free, autonomous robot (left) and the microfluidic device that controls it (right). Sources: https://wyss.harvard.edu/news/the-first-autonomous-entirely-soft-robot/ | https://newatlas.com/ chemical-power-soft-robot-autnomous-harvard/45073/ properties, have difficulty with it or requiring excessive computational resources. Soft robot control Soft robots can be controlled by various means, such as embedded traditional control using a CPU, artificial intelligence (AI) and machine learning (ML). AI/ML can also be used to process data from sensors to manage the control of soft robots, enabling autonomous and adaptive behaviours. Integrated sensing and feedback loops using soft, stretchable sensors enable soft robots to perceive their own shape by detecting pressure and touch. Microfluidics is also used to produce simple, repetitive motions, but not complex decision making and control as with a CPU and AI. Traditional control systems are well known, so we will just discuss microfluidic controls in detail. Here are some examples: Conventional computing platforms, such as Arduinos, can be converted into stretchable, compliant controllers by embedding them in flexible carriers with highly stretchable conductors, as demonstrated by researchers at the Yale University Faboratory (see Fig.11). The team created stretchable versions of Arduino Pro Mini boards that function at over 300% strain, embedding them directly into soft robots for locomotion control and wearables for motion sensing. The conductors are made from biphasic gallium-indium alloys, particularly oxidised gallium-indium (OGaIn), a foam of amorphous gallium oxide particles mixed with EGaIn. It was patterned on or within silicone substrates for high conductivity, extreme stretchability and reliable interfaces with rigid components. For more on this, see the video at https://youtu.be/VgNwUPpOY9A We also looked at flexible electronics in our November 2015 issue. Microfluidics; the main purpose of microfluidics in soft robotics is to enable autonomous, electronics-free (or minimal-electronics) control of soft robotic systems, particularly for untethered and lightweight designs. Microfluidics involves routing small volumes of fluids (gases or liquids) through tiny embedded channels and valves within the soft robot’s body. These channels form fluidic logic circuits, analogous to electronic circuits that can perform basic computation (eg, AND/OR/NOT gates, oscillators, timers) using pressure differences instead of electricity. Fig.11: an Arduino microcontroller module with the components mounted on a stretchable substrate with flexible conductors. Source: https://engineering.yale. edu/news-and-events/news/flexible-electronics-stretching-possibilities-softrobots siliconchip.com.au Australia's electronics magazine Microfluidic controllers contain components like pumps, fluid logic gates analogous to transistors, oscillators, shift registers, multiplexers and fluidic amplifiers. Microfluidics is good for generating rhythmic or sequential motion, reducing weight and complexity. It also enables some degree of untethered autonomy because the fluidic controllers can be powered by onboard chemical reactions or stored pressurised gas, allowing operation in environments where electronics would fail (eg, underwater, in MRI machines or explosive areas). Also, distributed control is possible as pressure signals propagate through the body like nerves, enabling coordinated multi-limb movement without a central processor. Microfluidic logic excels at simple, repetitive tasks (eg, walking gaits, pulsing, or basic sensing feedback) but not complex tasks. It is therefore most valuable in minimalist untethered robots or as a low-level controller complemented by higher-level electronics in hybrid designs. It serves as the brain and nervous system for the simplest fully soft, autonomous robots, trading computational power for lightness, robustness and independence from external power/control tethers. An example of a microfluidic controller is the one used in the Octobot (Fig.12; described in more detail later). Its controller (Fig.13) is basically an oscillator circuit. It was created using soft lithography, in which PDMS (polydimethylsiloxane) is poured into moulds with etched channels. The polymer was cured and solidified, with multiple layers being precisely aligned and bonded to form July 2026  17 Fig.14: in a Quake valve, the control air can deform the flexible membrane and block the flow of the process fluid. Fig.15: a pneumatic ring oscillator. Source: https://is.mpg.de/ publications/preston19-scir-oscillator Fig.16: the Squishy Robotics mobile robot. Source: https://squishy-robotics.com/ research-and-development/ the fluidic valves, channels and oscillator network. The Quake valve, invented by Stephen Quake and collaborators in 2000, is a widely used pneumatic microfluidic valve in soft robotics and lab-ona-chip devices. It typically consists of multi-layer soft elastomer structures with two crossing channels: a flow channel (for the main fluid/gas) and a perpendicular control channel (for pressurised air) – see Fig.14. When pressure is applied to the control channel, a thin elastomeric membrane between the channels deflects and pinches off the flow channel, closing the valve (acting like a transistor for fluid flow). Releasing pressure opens it again. This design can be used to make fluidic logic gates (AND, OR, NOT, oscillators etc) by combining multiple valves, enabling complex control without electronics, a key for untethered soft robots It is scalable (thousands can be integrated on one chip), has a fast response time and is compatible with pneumatic actuators. A disadvantage is that it is limited to low flow rates, so it is often better for microscale or control signals rather than high-force actuation. Microfluidic pumps are used to pump fluids around the microfluidic chip and through devices like the Quake valve shown in Fig.14. Each tube is activated in sequence to push fluid along in a peristaltic fashion. Microfluidic ring oscillators are pneumatic (or hydraulic) devices that offer a clever way to generate periodic motion in soft robots without any electronic control – see Fig.15. A ring oscillator consists of an odd number (typically three or five) of pneumatic inverters, which are fluidic valves made from deformable elastomeric membranes that function like inverting logic gates with hysteresis, similar to a Schmitt trigger inverter. When connected in a closed loop, pressure builds sequentially in each stage, causing the inverters to switch between on and off states. This creates a self-sustaining oscillation that can drive rhythmic movements, such as the coordinated leg motion in multi-legged crawling robots. These fully soft oscillators enable truly untethered operation, as demonstrated in some autonomous soft walkers such as the UC San Diego 18 Silicon Chip Australia's electronics magazine soft-legged walking robot described later and shown in Fig.19. Advantages of soft robots The advantages include: • Superior safety in human-robot interaction due to compliant, deformable bodies, making them inherently safe for close collaboration with people (eg, in factories or assistive wearables). • Adaptability to unstructured environments, being able to squeeze through narrow gaps, navigate irregular terrain or handle objects of varying shapes and fragility without precise programming or complex sensors. • Biomimicry and natural movement; some soft robots replicate biological systems (octopus tentacles, elephant trunks, worms), enabling smooth, energy-efficient, and fluid motion that rigid robots struggle to achieve. • Robustness to impacts and overload; compliant materials distribute forces, allowing them to survive drops, collisions or compression (eg, Squishy Robotics’ airdropped platforms). • They are lightweight and potentially low-cost because many use inexpensive elastomers and simple siliconchip.com.au fabrication methods (3D printing, moulding), reducing overall system weight and enabling untethered operation in some cases. issue (siliconchip.au/Article/17782). It produced the ReStore Exo-Suit, designed to promote the redevelopment of correct gait patterns in people who have suffered a stroke or similar Disadvantages of soft robots neurological injury. The disadvantages include: It is a lightweight, cable-driven soft • Limited force and precision; soft robotic device that provides timed materials generally produce lower assistive forces to a partially paralysed forces and less precise positioning ankle and is commercially used in compared to rigid actuators and metal rehabilitation clinics. It uses Bowden linkages, making them unsuitable for cables (like bicycle brake cables) heavy lifting or high-accuracy tasks. pulled by motors in a waist-mounted • Durability and fatigue; repeated unit. For more details, see the video at deformation causes material fatigue, https://youtu.be/1pC3fUGOdFw wear, tearing or degradation over time, Somnox (https://somnox.com) is especially under high strains or cyclic a pillow-like robot that simulates loading. breathing patterns to help with sleep • Complex modelling and control; disorders. It is pneumatically opertheir nonlinear, infinite-degree-of-­ ated. According to the manufacturer, freedom deformation makes accurate “Somnox detects and matches your modelling, simulation and precise breathing rate and rhythm; then, the control difficult. rhythm gradually slows to a tranquil, • Power and actuation challenges; sleep-inducing pace”. many rely on bulky external compressors (for pneumatics/hydraulics) or Rescue robots high voltages (for dielectric elastoSquishy Robotics (https:// mers), limiting true untethered perfor- squishy-robotics.com) produces both mance. Onboard chemical or battery a commercial stationary soft robot and solutions are still emerging. a developmental mobile one, shown • Manufacturing and scalabil- in Fig.16. The stationary robots can be ity; while prototyping is relatively deployed from aircraft to be dropped easy, producing durable, repeatable, into disaster areas from 300m and can high-performance soft robots at scale carry a variety of payloads such as gas remains challenging and expensive sensors for CO, H2S, lower explosive compared to conventional robotics. limit of gas and O2, cameras for 360° video, GPS and mesh networking. Commercial applications The structural concept is tensegrity (tension and integrity), which uses a network of rigid struts in compression and elastic cables in tension to create compliant, lightweight robots that are deformable and can absorb impacts such as being dropped from an aircraft. The developmental mobile robot moves via the concept of paired cable actuators, where one cable pulls and the other pushes like pairs of muscles work in opposition. In these robots, there are multiple sets of paired actuators acting in a coordinated way to propel the robot, as shown in the video at https://youtu.be/oDaLb63iCPI Soft grippers and food handling Schmalz Group (www.schmalz. com) produces the mGrip range of pneumatically powered soft robotic grippers for picking up fragile items like produce, baked goods and poultry without damage (see Fig.17). The grippers are activated by air channels inside the ‘fingers’. They are an example of a soft robotic component attached to a traditional rigid robot. The Festo HPSX (https://press.festo. com/en/node/5135) silicone gripper is a soft robot component for high-speed picking up of delicate products such as various foods – see Fig.18. It can work safely in human-robot collaborative environments with low risk to humans. It is pneumatically operated. Soft robots excel where safety, adaptability and gentle interaction are priorities, such as in medical devices, food handling or exploration. Still, they lag behind rigid robots in strength, precision and long-term reliability. Most practical applications today use hybrid approaches, combining soft elements for interaction with rigid frameworks for power and control. Here are some examples. Healthcare and emotional support Moflin from Casio (www.casio.com/ us/moflin) is an AI fluffy companion for emotional support that develops personality traits. It is regarded as a soft robot by some, but while its exterior is soft and fluffy, its interior is standard mechatronics. ReWalk Robotics (now Lifeward) produced the walking assistance exoskeleton described in our article on prosthetic limbs in the March 2025 siliconchip.com.au Fig.17: the mGrip soft robotic gripper. Source: Schmalz – siliconchip.au/ link/acaw Australia's electronics magazine Fig.18: the Festo HPSX silicone soft robot gripper. July 2026  19 iCobots (https://icobots.com) is an Israeli company providing plug-andplay soft robotic grippers that integrate seamlessly with existing industrial robots and cobots. A cobot is a collaborative robot designed to work alongside people rather than replace them. These grippers combine the speed of automation with the gentle, adaptive touch of human hands, making them ideal for handling delicate items such as eggs, fruit, chocolate and other fragile produce without damage or the need for complex vision systems. Wearable and assistive devices NEO from 1X Technologies (www. 1x.tech/neo) is a domestic humanoid robot with a soft-bodied design, launched in late 2025 and now available for pre-order. It is intended for everyday household tasks and uses biology-inspired tendon-driven actuation (with high-torque-density motors) for smooth, compliant and safe interactions with people. Conceptual soft robots (experimental) Apart from commercial soft robots, many have been designed or are being researched as in the following: • In 2016, the Wyss Institute at Harvard University unveiled the first entirely soft autonomous robot, called Octobot. It was electronics-free, mostly 3D printed and used microfluidic logic oscillator circuits for control. It used hydrogen peroxide as fuel, which generated gas to drive the robot – see Fig.12 and the video at https://vimeo. com/179510230 • Stanford University has developed an “isoperimetric” soft robot in the form of an inflatable tube truss with relocatable joints which enable it to change shape due to changes in the length of individual truss members and move or perform other tasks (siliconchip.au/link/acap). The joints are moved by roller modules that create new joints by pinching the tube at different locations. See the video at https://youtu.be/XqgbLb8m77U • Researchers at UC San Diego have developed an electronics-free softlegged walking robot (Fig.19). It is powered by pressurised air and has no electronics. Its movement is controlled by pneumatic ring oscillator fluidic control circuits to generate rhythmic movement, similar to animals. The robot was also equipped with simple sensors in the form of bubbles of fluid at the end of the legs which, when depressed, flip a valve and cause the robot to change direction in response to environmental interactions. The biological inspiration for this machine comes from the African sideneck turtle. For more, see the video at https://youtu.be/bnT6BBkDYlc • Researchers at the University of California, San Diego, have developed an electronics-free autonomous walking robot with an embedded pneumatic oscillating control circuit. After 3D-printing the six-legged robot, shown in Figs.20 & 21, is ready to operate as soon as a gas supply (CO2 cylinder, tube and pressure regulator) is added. It uses a 3D-printable four-phase bistable oscillating valve, capable of generating coordinated motion of multiple limbs from a steady source of gas. Each of the six legs has four chambers, each of which generates one of up, down, forward or backward motion. See the videos at https://youtu.be/ f8hTK7AabM8 and https://youtu.be/ PDoiguTdLXs • SoFi is a soft robotic fish developed by MIT’s CSAIL in 2018. It is a remote-controlled underwater robot equipped with a camera to observe marine life without disturbing it. A diver directs it from a console (using acoustic signals) while hydraulic fluidic actuators in the tail mimic natural fish swimming. See Fig.22 and the videos at https://youtu.be/BSA_zb1ajes and https://youtu.be/Dy5ZETdaC9k • In recent years, Chinese researchers have pioneered soft robots for exploring the Mariana Trench, the ocean’s deepest point. Unlike traditional submersibles made of expensive Fig.20: a six-legged walking robot that needs no electronics; just a CO2 canister. Source: UC San Diego – siliconchip.au/link/acax Fig.19: the soft-legged walking robot from UC San Diego. Source: https:// newatlas.com/robotics/air-powered-robot-no-electronics-turtle/ 20 Silicon Chip Australia's electronics magazine Fig.21: the embedded pneumatic oscillating control circuit of the robot shown in Fig.20. siliconchip.com.au Fig.22: a SoFi robotic fish. Source: MIT News – siliconchip.au/link/acay Fig.23: Zhejiang University’s 2021 robot compared to a snailfish. Source: www. zju.edu.cn/english/2021/0317/c65148a2268191/page.psp hard metal shells to resist the extreme pressures at depth, these robots are soft, and external pressure is distributed evenly throughout them, just as in the fish they mimic. A landmark 2021 design from Zhejiang University mimicked the hadal snailfish, using a silicone body (22cm long with a 28cm fin span) and dielectric elastomer actuators for a flapping motion. It reached a depth of 10,900m, a world record for a soft robot – see Fig.23. Building on this, a 2025 robot from Beihang University (inspired by batfish locomotion) reached 10,666m, enduring 1100 bar. This larger version (50cm long and weighing 2.7 kg) uses shape-memory alloy actuators that oscillate with periodic heating for multimodal movement of swimming, gliding and crawling across the seafloor. • The DARPA ChemBot (Chemical Robots) program was a research initiative launched around 2007-2008 to develop soft, flexible, shape-shifting robots capable of squeezing through tiny openings (smaller than their normal size), reconstituting their shape and regaining function on the other side, to perform tasks like reconnaissance or payload delivery in denied/ hostile environments. The research program finished in 2011-2012 with no further announcements. • The Amphibious Robotic Turtle (ART), developed by Yale University researchers (Figs.24 & 25) is a bio-­ inspired soft robotic platform with a solid body but soft robotic legs that employs ‘adaptive morphogenesis’ to dynamically adapt its limbs for multi-environment locomotion. Its cylindrical legs can morph into flattened flippers for efficient swimming in water, mimicking sea turtles while reverting to load-bearing legs for land travel like tortoises. This transformation takes 1-2 minutes and is achieved using a thermally responsive polymer composite that softens when heated (via embedded heaters) and stiffens when cooled to hold the new shape. An internal soft pneumatic ‘muscle’ (balloon-like structure) inflates or deflates to drive the shape change during the malleable phase, enabling seamless transitions between terrestrial gaits (creep/crawl) and aquatic propulsion (flapping/paddling). • A survivable amputation of a body part to escape danger is a survival strategy used by certain lower animals like lizards, starfish and crabs. A soft robot has been developed at Yale University to do the same thing. For example, if the leg of a search and rescue robot gets trapped by falling debris, a built-in heating element can melt it away. See the video at https://youtu. be/qPd9x9-bALo • Harvard University’s Whitesides Research Group has developed the “arthrobot”, with an exoskeleton constructed from thin polymeric tubes. It also has pneumatic joints modelled after the hydrostatic joints of spiders to provide actuation and mechanical compliance to external forces. An inflatable elastomeric tube extends a Figs.24 & 25: an amphibious robotic turtle. Source: https://yaledailynews.com/blog/2022/10/25/yale-led-team-developsshape-shifting-turtle-robot/ siliconchip.com.au Australia's electronics magazine July 2026  21 Fig.26: Harvard University’s Whitesides Research Group arthrobot. Source: www.gmwgroup.harvard. edu/soft-robotics limb while an opposing elastic tendon retracts it – see Fig.26. Experimental grippers • The Festo Bionic Learning Network Octopus Gripper uses pneumatic tentacles and vacuum suction to hold objects of any shape. It is currently not a commercial product, but aspects of its technology have been incorporated into other Festo products. See Fig.27 and the video at https://youtu. be/w1zU7FNKm_w • The University of California, San Diego (UCSD) has developed a 3D-printed (in one print) gripper that has an embedded microfluidic controller and needs no electronics to operate. When the gripper is moved horizontally, it drops the object. See Fig.28: the UCSD no-electronics gripper. Source: Iguana Robot – siliconchip.au/link/acaz 22 Silicon Chip Fig.27: the Festo Octopus Gripper, also known as the TentacleGripper. Source: www.festo.com/gb/en/e/about-festo/research-and-development/bioniclearning-network/bionic-grippers-and-soft-robots/tentaclegripper-id_33321/ Fig.28 and the video at https://youtu. be/A5mpy3X1dcc • A granular jammer is a soft robotics gripper concept where grains or grain-like materials are placed inside a membrane, such as a balloon. It is placed around an irregularly shaped object and then a vacuum is applied to tighten the grip. The object can then be moved – see Fig.29. • The Jamming Donut is a universal soft-robotics gripper designed by Australia’s CSIRO. This doughnut-shaped gripper can grab round objects like doorknobs. • A team at the University of North Carolina at Chapel Hill has developed soft robots made of two layers, one simulating skin and the other muscle, that can autonomously detect and respond to different physiological stimuli. The robot’s base layer is made from a thermally responsive hydrogel that can contract and relax like muscle, allowing the robot to bend. The other layer is an electronic ‘skin’ made of another soft polymer, which can host a variety of sensors or stimulators. Such sensors can detect acidity, electrical activity, mechanical strain and temperature; mini electrodes could stimulate tissue, while electrical heaters could trigger the robot’s hydrogel ‘muscle’ layer to contract – see Fig.30. Onboard electronics allow for wireless power and data transmission. Fig.29: a granular jammer or “jamming gripper”. Source: www. creativemachineslab.com/jamminggripper.html Fig.30: sensors from an experimental soft robot. Source: www.nibib.nih.gov/ news-events/newsroom/taking-cuesnature-medical-soft-robots-get-smart Experimental medical applications Australia's electronics magazine siliconchip.com.au (A) (B) (C) Fig.33: the concept of Vine Robot’s movement. It lengthens and doesn’t slide, thus it has no problem getting through small openings. Fig.34: the principle of steering and extension of a vine robot. (a) Air pressure is applied to the core. (b) the pressure causes extension. (c) differential pressure is used for steering. Source: www.researchgate.net/publication/368389948 • Harvard University’s Wyss Institute has developed a soft-robotic sleeve that uses pneumatic actuators and wraps around a human heart to assist its beating. As there is no direct contact with blood, the patient does not have to take blood thinners, as with conventional artificial hearts or ventricular assist devices. The device is currently in advanced preclinical testing stages – see Fig.31. • Fig.32 shows a soft prosthetic robotic hand for amputees or service robots. It has the advantage of feeling more like a normal human hand. It was developed by Rob Scharff from the Delft University of Technology. • Researchers at AMOLF, a leading Dutch institute for fundamental physics and soft matter, have developed a remarkable soft robotic artificial heart prototype that demonstrates autonomous beating with minimal electronics. Driven entirely by pneumatic pressure from an external pump, the device uses advanced soft biomaterials and clever fluidic logic to produce a repetitive heartbeat. A key innovation is a passive soft valve inspired by the sputtering effect when squeezing an almost-empty plastic sauce bottle. As pressure builds, a soft tube buckles and rapidly opens/ closes, creating self-oscillating flow that drives rhythmic contraction of the heart without electronic controllers. • The DARPA Soft Exosuit is a wearable soft robot developed at the Wyss Institute at Harvard University. It works with the body’s muscles to reduce fatigue and assist movement. Cables and motors are used to apply forces to hips and ankles, mimicking tendons and muscles. It can be used in applications such as assisting stroke patients or soldiers. See the video at https://youtu.be/aeDm5yFYt10 Experimental soft robots that mimic plant tendrils These robots grow and navigate by extruding material or inflating tubes. Examples include Vine Robots and the FiloRobot. Vine Robots from Stanford University and UCSB, Dr Elliot Hawkes (www.vinerobots.org/about/) use soft pressurised polyethylene tubes that evert (turn inside out) to form a body that can navigate numerous types of obstacles, even a field of pointed nails or glued-together boards – see Figs.33, 34, 35 & 36. It does not slide; the body extends and lengthens from the tip. That means there is no friction between Fig.31: the Harvard cardiac assistance device. It’s a sleeve that wraps around the heart. Source: https://seas. harvard.edu/news/2017/01/soft-robot-helpsheart-beat Fig.32: a soft robotic prosthetic had for amputees or service robots. Source: https://elveflow.com/ microfluidic-reviews/soft-robot/ siliconchip.com.au Australia's electronics magazine July 2026  23 – see Fig.38 and the video at https:// youtu.be/e1mOac3wRsw Experimental photoresponsive soft robots Fig.35. Vine Robot navigates through a hole. Source: ExtremeTech – siliconchip.au/link/acb0 Fig.36: Vine Robot navigates a maze. Source: www.science.org/doi/10.1126/ scirobotics.aan3028 the external body of the robot and the surfaces it contacts. It can reach up to 72m from its base. Fig.34 shows the principle of extending and steering a Vine Robot. In (a), air pressure is applied to the core. In (b), the air pressure causes extension due to eversion of the tube. The outside of the tube doesn’t move with respect to the surface it contacts. In (c), the end is steered by varying the air pressure by applying air pressure to one or two of the serial pneumatic actuator muscles (sPAM) mounted around the robot’s circumference at the tip of the robot. Guidance is through the use of an inertial measurement unit (IMU) at the tip and a shaft encoder at the base to sense the amount of extension. There is also a camera at the tip. For more information, see the video at https:// youtu.be/q2Q-taHAo7Q The Vine Robot is not commercially available, but has inspired spinoffs using the eversion technology and a robotic gripper for industrial and elder care used to assist in lifting objects or people from beds; see Fig.37. Systems are also being tested for the inspection of commercial pipe networks. There are instructions to build your own Vine Robot at www.vinerobots. org/build-one Miniaturised 1.8mm-diameter versions are also being developed for non-invasive surgeries and intubation – see https://pmc.ncbi.nlm.nih. gov/articles/PMC12370164/ Filobot from the Istituto Italiano di Tecnologia (IIT) is an experimental robot and is not strictly a soft robot, but it bears some resemblance to the Vine Robot. It prints its own stem by 3D-printing its own body from the tip as it ‘grows’ Fig.37: a Vine Robot inspired gripper from MIT. Source: https://news.mit. edu/2025/vine-inspired-robotic-gripper-gently-lifts-heavy-and-fragileobjects-1210 24 Silicon Chip Australia's electronics magazine The Max Planck Institute in Germany has developed untethered soft micro-robots that walk or roll toward/ away from light sources (phototaxis). The Chinese Academy of Sciences has developed near-infrared-driven soft grippers for remote manipulation in confined spaces. Conclusion Soft robots represent a shift toward more organic, resilient machines, with the potential to revolutionise fields from healthcare to exploration, although challenges remain. More videos If you’re interested in seeing more about soft robots, then check out these videos below: “DIY a Food-Grade Soft Gripper for Your Delta X S Robot for Just... $10”: https://youtu.be/Eo3y-UqJ100 “DIY Soft Robotic Tentacle”: https://youtu.be/gPYjo-W2ctU “Can You 3D Print a Robot’s Brain Out of Air?”: https://youtu.be/Cn7jC6YamGE “I Printed a Microchip That Runs on Air — A Nervous System for Squishy Robots!’: https://youtu.be/QJdBp5dGrww SC Fig.38: FiloBot climbs a tree. Credit: Del Dottore et al., Sci. Robot. 9, eadi5908 (2024) siliconchip.com.au