Soft Robot Design with BiomimeticsFollow article
What soft robots lack in hard strength and precision, they make up for with their gentle dexterity handling delicate objects, or their ability to go places other robots cannot reach.
SoFi the swimming robot Picture credit: CSAIL MIT
Soft Robots and Cobots
Back in 2013, I wrote an article for DesignSpark entitled Soft Humans, Hard Robots which introduced the concept of ‘soft robotics’, a new area of research which was rapidly becoming mainstream. At that time, most of the activity involved developing actuators such as pneumatic muscles and grippers able to pick up fragile objects without breaking them – just like a human hand. The term Cobots nowadays applies to robots with a ‘human-friendly’ construction and safety attitude, able to work uncaged alongside human workers at a bench. Where the Cobot uses careful design and soft materials to pad out its hard internal structure, soft robots are often made entirely of compliant materials, not to make them safe with humans, but to enable them to work in decidedly hazardous or awkward environments. In biological terms, the cobot is a vertebrate with an internal ‘bone’ structure giving it strength and the capability of precision movement. The soft robot is usually an invertebrate, like a jellyfish or octopus, able to squeeze through narrow gaps, bend around uneven terrain or grip irregular objects. In fact, Mother Nature is a favourite source of ideas for soft robotic mechanisms. This general field of study goes by the name of Biomimicry or Biomimetics.
Biomimetics has many different definitions depending on the exact application area. For soft robotics, it can be described as taking inspiration from the function of a biological mechanism, for example, the human hand, and creating a functionally similar artificial version – a robotic gripper. The key phrase is functionally similar, not an exact copy. Another example would be an aeroplane wing: inspired by birds’ wings but using different materials. This paper by George Whitesides, ‘Bioinspiration: something for everyone’ and the following video lecture explain the thinking behind biomimetics in soft robot design.
Soft robots are still very much in the development stage: you won’t find many commercial products for sale at the moment. As a result, this is very much an R & D focused article with video clips from and links to university researchers’ work. Most development takes place in four overlapping areas; grasping irregular and/or fragile objects, sensing, artificial muscles, and propulsion. Strictly speaking, a soft robot ought to be made entirely of compliant materials (no hard bits like PCB-based electronics), but there are ‘grey areas’. For example, a conventional jointed arm might have a soft gripper for handling delicate parts.
As most small robotic arms spend their lives picking things up and putting them down somewhere else, the most common ‘end-effector’ in use is the gripper. A ‘hard’ robotic gripper is rigid and can usually apply sufficient force to crush a fragile object. Without any tactile feedback or object recognition, it’s up to the human operator/programmer to decide how much force is sufficient to grip without breaking the object. A very difficult task to automate. Soft grippers based on pneumatics or electrostatic adhesion are superior in a variety of ways: they are very gentle, most can handle objects of different shapes and weights, and control is much simpler without the need for tactile feedback. Take a look at this engineering prototype from Harvard University for a robot arm and gripper based on an octopus tentacle:
This device uses compressed air to move the arm with a vacuum applied to grip the object through the ‘suckers’. There is nothing new in using pneumatics or hydraulics to drive a robot arm. What is new, is the use of biomimicry in the design of the actuators. Traditional piston-in-cylinders (rams) have been replaced by cellular structures which can produce the exact movement required directly without levers, cams and complex control systems. Even rotary motion can be achieved, as we shall see later. The next video shows an electrostatic gripper that can pick up just about anything:
The humanoid robot lies in that grey area between soft and hard robotics, because of its load-carrying ‘bone’ structure. Early attempts to make full-size ‘androids’ used pneumatic/hydraulic rams to simulate muscle action. This was never very successful because of the weight of these things and the large number required to mimic even basic human locomotion. A lot of work is going on developing lighter, more flexible actuators. In this video, novel constructional methods using standard materials yield pneumatically-powered soft ‘muscles’:
An example of soft robot design being precisely dictated by the application is described in the paper: ‘A Soft Tube-Climbing Robot’. The entire robot is just pneumatic muscle! Researchers are looking at producing new materials that expand or contract when heat or an electric current is applied. This paper titled ‘Soft material for soft actuators’, describes a special polymer infused with liquid ethanol bubbles. An embedded wire is heated by a low-voltage electric current, causing the ‘muscle’ material to expand.
How about this octopus robot (octobot) from Harvard University? Built on a 3D printer it uses gas pressure to flex its tentacles. The gas comes from a catalytic reaction between platinum and hydrogen peroxide, otherwise known as diluted rocket fuel. The video shows how it’s made:
The electrostatic gripper described above is an example of a Dielectric Elastomer Actuator (DEA). Its principle of operation can be adapted to create an artificial muscle for other purposes. A novel variation is described in the paper titled: ‘Translucent soft robots driven by frameless fluid electrode dielectric elastomer actuators’. This muscle is designed for use underwater with one electrode being the sea itself. It’s able to reproduce the slow undulating motion of a fish or eel body. In this case, the inspiration came from a particular eel larva, leptocephalus. A useful feature of this robot eel is it’s near transparency; seeming to vanish when placed in the water.
The artificial muscles I’ve been talking about would provide the necessary movement for any kind of walking robot, humanoid or multi-legged. What about driving wheels? The same principles of inflating cells used in grippers can be used to drive a wheel. Rotary actuators are demonstrated in this video of an entirely ‘soft’ wheeled-rover:
An octopus uses jet propulsion for high-speed motion. That in itself is not very interesting. What attracted researchers is the way it does it. In principle, a sudden muscular contraction squeezes a bladder full of water, so generating the ‘rocket’ effect. In the paper: ’Ultra-fast escape manoeuvre of an octopus-inspired robot’ they describe how as the bladder deflates, the octopus becomes more streamlined. As the jet peters out, the better hydrodynamics partly compensate for the loss of thrust. These creatures provide all sorts of biomimetic inspiration as seen in the introductory video with even more ideas in the final video of this article.
The next two research papers show how the natural world can provide a fresh angle on sensor design. The first, ‘Bioinspired polarization vision enables underwater geolocalization’ demonstrates the use of light polarisation-sensitive vision of underwater sunlight to determine position – a low-precision GPS. The bio-inspiration came from studying the eyes of shrimps.
The second, ‘Fully 3D Printed Multi-Material Soft Bio-Inspired Whisker Sensor for Underwater-Induced Vortex Detection’ describes a highly specialised underwater sensor. The whiskers of a common seal detect the vortices produced in the water by its swimming fish prey. The researchers developed an artificial version for a soft robot fish.
Most soft robots have at least one bio-inspired feature and soft grippers are likely to be must-have options with Cobot arms soon. This final video contains more examples of technology transfer from the natural to the artificial world featuring once again, the versatile octopus.
If you're stuck for something to do, follow my posts on Twitter. I link to interesting articles on new electronics and related technologies, retweeting posts I spot about robots, space exploration and other issues.