What makes a robot soft
Most robots are built like machines: rigid links joined by motors, with sensors and a computer deciding how to move them. A soft robot is built mainly from materials that bend, stretch or squash, such as silicone rubber, fabric, gels or flexible plastics. It typically moves by changing the pressure of air or fluid in internal chambers, or by using materials that change shape under heat or electricity. Because the body itself is compliant, it can squeeze through gaps, adapt around an object it grips, and bump into things with less risk of damage.
The field is usually dated to the mid-2000s and the years that followed, but it was assembled from older ideas. This article covers soft robots in general. Soft wearable suits are a specialized application with their own history.
Earlier ancestors
Several earlier lines of work anticipated the field. Pneumatic artificial muscles from the 1950s showed that a rubber tube could contract when inflated. In the 1970s the Japanese engineer Shigeo Hirose developed snake-like robots and other designs inspired by animals, which showed how bodies without legs or wheels could move. Researchers in the 1980s and 1990s also built continuum arms and tentacle-like manipulators, often inspired by elephant trunks and octopus arms, that bent smoothly instead of at discrete joints.
Meanwhile, work on robot hands recognized that a gripper which complied with an object was more forgiving than a rigid pincer that had to be positioned exactly. This “compliance” idea, whether achieved by springs, rubber pads or air pressure, became central.
Becoming a field
By the late 2000s, several groups were using the term “soft robotics” and sharing methods. A well-known contribution was the work of George Whitesides’s group at Harvard, which popularized a way of casting silicone rubber in molds with internal air channels so that inflating them made the body curl or crawl. A widely noted 2011 paper described a soft robot that could walk and squeeze under obstacles, using only changes in pressure. The materials were cheap and the molding process was simple enough that many laboratories could join in, which helped the field grow quickly.
Some other landmarks are:
- A “jamming” gripper, reported around 2010 by teams from Cornell, the University of Chicago and iRobot, which used a soft bag of granular material, such as coffee grounds, that stiffens around an object when air is pumped out.
- Octopus-inspired arms and robots that use fabric and cables.
- In 2016 the Harvard group introduced a small, entirely soft “octobot” that was powered by a chemical reaction. It was a demonstration of what might be possible rather than a practical machine.
An international conference series dedicated to soft robotics began later in the 2010s, a sign that the field had become established.
Making soft robots
Manufacturing is a recurring theme. Molding and casting were the first methods. Later researchers used additive manufacturing to print flexible bodies with internal channels in one piece, and rapid prototyping workflows are common in the field, because designs are tested and discarded quickly. Printing and molding help, though sealing, durability and repeatability remain problems for soft parts that are cycled thousands of times.
Control is also different. A rigid arm has joints whose angles can be measured and commanded. A soft body can deform in many ways at once, and its sensors, such as stretchable strain gauges, may be less precise. As a result, many soft robots rely on carefully designed bodies that do part of the “thinking” through their shape, rather than on complex software.
Early uses
The earliest practical uses were generally in grasping. Soft gripper fingers are now used in some food-handling and packaging systems, where items such as fruit or baked goods are fragile or irregular in shape. Soft robots have also been explored for inspecting tight spaces, for minimally invasive medical tools such as flexible catheters, and for underwater work where gentle contact with delicate organisms is useful. Many of these are still at the research or pilot stage. Medical applications in particular require extensive testing, and any account of them here is historical only.
Limits
Soft robots are not a replacement for conventional ones. They are generally weaker, slower and harder to control accurately, and they often need pumps or compressors that are bulky. Their strength is in safe contact and adaptability rather than power. The field’s history shows the same pattern as other kinds of robotics: once the basic idea was available, progress depended on materials, manufacturing methods and practical applications catching up.