Spiders are truly amazing creatures. They have evolved over more than 200 million years and can be found in almost every corner of our planet. They are one of the most successful animals. Not less impressive are their webs, highly intricate structures that have been optimised through evolution over approximately 100 million years with the ultimate purpose of catching prey.

However, interestingly, the closer you look at spiders’ webs the more details you can observe and the structures are much more complicated than one would expect from a simple snare. They are made of a variety of different types of silks, use water droplets to keep the tension [see: citation 4], and the structure is highly dynamic [see: citation 4]. Spider’s webs have a great deal more morphological complexity than what you would need to simply catch flies.

Since nature typically does not spoil resources the question arises: why are spiders’ webs so complex? Might they have other functionalities besides being a simple trap? One of the most interesting answers to this question is that spiders might use their webs as computational devices.

How does the spider use the web as a computer?

Despite the fact that most spiders have a lot of eyes (the majority has 8, but some have even up to 12), a lot of the spiders have bad sight. In order to understand what is going on in their webs, they use mechanoreceptors in their legs (called lyriforms) to “listen” to vibrations in the web.  Different species of spiders have different preferred places to sit and observe. While some can be found right at the center, others prefer to sit outside the actual web and to listen to one single thread. It is quite remarkable that based only on the information that comes through this single thread the spider seems to be able to deduce what is going on in their web and where these events are taking place.

For example, they need to know if there is a prey, like a fly, entangled in their web. Or if the vibrations are coming from a dangerous insect like a wasp and they should stay away. The web is also used to communicate with potential mates and the spider even excites the web and listens to the echo. This might be a way for the spider to check if threads are broken or if the tension in the web has to be increased.

From a computational point of view, the spider needs to classify different vibration pattern (e.g., prey vs predator vs mate) and to locate its origin (i.e., where the vibration started).

One way to understand how a spider’s web could help to carry out this computational functionality is the concept of morphological computation. This is a term that describes the understanding that mechanical structures all over in nature are carrying out useful computations. For example, they help to stabilise running, facilitate sensory data processing, and helps animals and plants to interact with complex and unpredictable environments.

One could say computation is outsourced to the physical body (e.g., from the brain to another part of the body).

From this point of view, the spider’s web can be seen as a nonlinear, dynamic filter. It can be understood as some kind of pre-processing unit that makes it easier for the animal to interpret the vibration signals. The web’s dynamic properties and its complex morphological structure mix vibration signals in a nonlinear fashion. It even has some memory. This can be easily seen by pinching the web. It responds with vibrations for some moments after the impact echoing the original input.  The web can also damp unwanted frequencies, which is crucial to get rid of noise. On the other hand, it might even be able to highlight other signals at certain frequencies that carry more relevant information about the events taking place on the web.

These are all useful computations and they make it easier for the spider to “read” and understand the vibration patterns. As a result, the brain of the animal has to do less work and it can concentrate on other tasks. In effect, the spider seems to devolve computation to the web.  This might be also the reason why spiders tend to their webs so intensively. They constantly observe it and adapt the tension if it has changed, e.g. due to change in humidity, and repair it as soon a thread is broken.

From spider webs to sensors

People have speculated for a while that spider webs might have additional functionalities. A great article that discusses that is “The Thoughts of a Spiderweb“.

However, nobody so far has systematically looked into the actual computational capabilities of the web. This is about to change. We recently started a Leverhulme Trust Research project that will investigate naturally spun spider webs of different species to understand how and which computing might take place in these structures.  Moreover, the project does not only try to understand the underlying computational principles but will also develop morphological computation-based sensor technology to measure flow and vibrations.

The project combines our research expertise in Morphological Computation at the University of Bristol and the expertise on spider webs at the Silk Group in Oxford.

In experimental setups we will use solenoids and laser Doppler vibrometers to measure vibrations in the web with very high precision. The goal is to understand how computation is carried out. We will systematically investigate how filtering capabilities, memory, and signal integration can happen in such structures. In parallel, we will develop a general simulation environment for vibrating structures. We will use this to ask specific questions about how different shapes and materials others than spider webs and silk can help to carry out computations. In addition, we will develop real prototypes of vibration and flow sensors, which will be inspired by these findings. It’s very likely they will look different from spider webs and they will use various types of materials.

Such sensors can be used in various applications. For example, morphological computation based flow sensors could be used to detect anomalies in the flow in tubes. Or vibration sensors put at strategic places on buildings could be able to detect earthquakes or structural failure. Also highly dynamic machines, for example, a wind turbine, could be monitored by such sensors to predict failure.

Ultimately, the project will provide not only a new technology to build sensors, but we hope also to get a fundamental understanding how spiders use their webs for computation.

References

[1] Hauser, H.; Ijspeert, A.; Füchslin, R.; Pfeifer, R. & Maass, W.”Towards a theoretical foundation for morphological computation with compliant bodies.”Biological Cybernetics, Springer Berlin / Heidelberg, 2011, 105, 355-370

[2]  Hauser, H.; Ijspeert, A.; Füchslin, R.; Pfeifer, R. & Maass, W. “The role of feedback in morphological computation with compliant bodies”. Biological Cybernetics, Springer Berlin / Heidelberg, 2012, 106, 595-613

[3] Hauser, H.; Füchslin, R.M.; Nakajima, K.“Morphological Computation – The Physical Body as a Computational Resource” Opinions and Outlooks on Morphological Computation, editors Hauser, H.; Füchslin, R.M. and Pfeifer, R., Chapter 20, pp 226-244,  2014, ISBN 978-3-033-04515-6

[4] Mortimer, B., Gordon, S. D., Holland, C., Siviour, C. R., Vollrath, F. and Windmill, J. F. C. (2014), The Speed of Sound in Silk: Linking Material Performance to Biological Function. Adv. Mater., 26: 5179–5183. doi:10.1002/adma.201401027

The Jaeger-Lipson coffee balloon gripper. Its soft end effector adapts itself to the object it grasps. Source: John Amend

Soft robots are versatile, often much safer, more energy-efficient, robust and resilient than their more rigid counterparts. But one of the biggest challenges facing soft robotics is control – often, classical approaches don’t apply. The answer may lie in morphological computation, an idea that stems from biological systems using their bodies to control basic actions.

Industrial Networking and Engagement Day on Soft Technologies

Explore the possibilities of soft technologies and soft robotics in your field!

October 8, 2015
Bristol, UK

Novel designs in soft robotics reference simplicity as well as versatility. The Jaeger-Lipson coffee-balloon gripper is one example. Essentially a balloon, it’s filled with ground coffee and can be either soft or hard depending on the air pressure within it. It can grasp objects using a simple one-dimensional control variable (switching the vacuum on or off). By exploiting its soft morphology, the end effector adapts itself to the object without the need for any explicit modelling or control.

Morphological computation proposes that at least some aspects of control (such as walking or grasping) can be outsourced to the body as these functions are already “encoded” within it. Another way to see it is that the body can be exploited as a computational resource. This makes a task much easier, since part of the “work” will already have been done by the body, reducing the complexity of the robot’s computational problems and the corresponding control and learning tasks.

The soft body of the octopus has potential as a computational resource.

So what kind of computation can we carry out in the physical bodies of robots? Biological systems have to take sensory information and interpret or transform it into sensible actuation signals. The computational process is a dynamic mapping of continuous input streams onto output streams, with a feedback loop through the environment. This could be, for example, a nonlinear controller that is implemented in the morphology to stabilize running – a controller is nothing other than a dynamical system, after all. In the case of morphological computation, this system is simply part of the body instead of an algorithm in some external, digital computational unit.

Fundamentally different from digital computing processes, morphological computation is not based on a Turing machine with discrete steps and finite time; it takes place in the continuous, analogue realm and is carried out directly in the body.

It can also be broadened out beyond the shape and form of body parts, into parameters that describe its dynamics, such as stiffness, friction, or damping. The dynamic parameters of the environment also shape the type of computation. For example, a compliant leg interacts with the ground (when it impacts), and a soft hand deforms by grasping an object.

Some background

Asimo has rigid body parts. While it is an ingenious piece of hardware, its body is not useful for morphological computation.

Recent studies in morphological computation [1][2] suggest that a computationally powerful body needs to exhibit highly complex dynamics (i.e. high-dimensional state space, nonlinearities and under-actuation). But in most classical robot designs, such as the Asmio robot and many industrial robots, these properties are deliberately suppressed as complex dynamics can interfere with modelling and control.

The concept of morphological computation has been widely used in robotics for some time now. Some nice examples are listed by Pfeifer and Bongard in their influential book “How the body shapes the way we think” [3].

Many of these robots were built using a combination of bio-inspiration, intuition, and clever engineering. Until recently, there were no clear guidelines for their design.

To fill this gap, we developed two theoretical, mathematical models [1],[2] that attempt to describe how bodies can be exploited as computational resources. In order to provide a descriptive generic model, we used networks of simple mass-spring damper systems – a good approximation for biological, but also soft, robotic bodies.

Caption: (a) The Cornell passive dynamic walker (Collins et al., 2001); (b) Wanda by Ziegler et al., 2006; (c) Puppy and (d) Stumpy by Fumiya Iida (Source: Iida and Pfeifer, 2004)

The underlying idea is to use morphology as a computational resource to emulate complex behaviours. The range of possible computation is impressive. Theoretically, there are almost no limits.

We were able to show that physical bodies can be used to emulate complex nonlinear differential equations (e.g. nonlinear controllers). We demonstrated setups to emulate nonlinear limit cycles that are highly robust and are even able to generalize beyond their learned data sets. Such limit cycles can be used to produce a control signal for locomotion. We were even able to demonstrate the use of the body as a switch between different limit cycles (behaviours) based on physical inputs (forces applied to the system) that could come from the environment. The implications are that the body can also serve as a type of sensor, enabling the robot to sense change through its soft body and react appropriately.

One possible application could be a robot that is able to “feel” the weight on its body, enabling it to adapt its locomotion by producing (computing) different limit cycles accordingly.

A robot that is able to “feel” the weight on its body and react appropriately by producing different limit cycles depending on the situation. Source: Hauser, H and Füchslin. RM and Nakajima, K. [4]

Building real robots

Theoretical results aside, a number of real-world platforms have exploited soft complex bodies as a computational resource. For example, in the EU project OCTOPUS we showed that a passive silicone arm (inspired by the octopus) can be used to carry out complex computational tasks [5 and 6].

Caption: (a) Passive soft silicone arm from [6]; (b) modular pneumatically driven robot arm, by TUM; (c) Kitty, a robot with compliant spine [7] (d) NASA’s superball robot [8]

We built a quadruped robot with compliant spine that we used as a computational resource to control its locomotion [7]. We also demonstrated, in a more industrial setting, that a modular, pneumatically-driven robot arm (developed at TU Munich by Martin Eder) can be exploited via morphological computation [paper close to submission]. Another example is NASA’s tensegrity robot “SuperBallBot,” with applications in exo-planetary exploration [8].

Implications and outlook

Morphological computation provides us with a new viewpoint: instead of perceiving their complex dynamics as a problem, we can instead embrace and exploit this hidden superpower of soft bodies. There are some interesting implications:

• Soft robots can exploit their bodies for computational tasks that can make them more robust, faster (they respond in real time) and, potentially, more energy efficient than their hard counterparts, as they work with rather than against their body dynamics.
• The addition of smart materials enables body parts to take over not only actuation, but also the task of sensing the environment and processing this information. This may even extend to situations where parts of the robot break down, resulting in highly resilient, adaptive and intelligent machines.
• A morphological computation approach will allow us to design robots that can easily switch from one behaviour to another. For example: from a stable limit cycle (during locomotion) to an exponentially stable equilibrium point (for a reaching task), e.g. [10].
• The concept can be applied to a range of intelligent structures. For example: smart furniture that adapts to the movement of its user, haptic interfaces and wearable soft devices for rehabilitation or teaching novel movements (e.g. a new sport). Even smart architecture that reacts to environmental changes.
• In the future, self-assembling systems could also exploit morphological computation. Since the morphology defines the function/computation, by changing the morphology, we can “reprogram” the machine.

Because of their complex dynamics, soft robots potentially have a more computationally powerful body than rigid robots. Morphological computation turns the complexity, which has been perceived as a disadvantage in robotics, into a powerful benefit. Looking at the work done so far, we believe there is great potential for other applications and novel ideas and we are looking forward to the future.

Want to learn more about soft robotics? Come to the  Industrial Networking and Engagement Day on Soft Technologies on October 8, 2015 in Bristol, UK. This is event is geared toward people from the private sector who want to explore the possibilities of soft technologies and soft robotics in their field.

If you liked this article, you may also be interested in Helmut Hauser’s presentation on Morphological Computation at the ShanghAI Lectures 2013. You may also enjoy:

• Soft robotics for adaptive building facades
• Two new soft robotics competitions announced by Soft Robotics Toolkit
• Robots Podcast: Soft Robotics Toolkit, with Donal Holland
• Robot Startup Series #7: Interview with Soft Robotics’ Carl Vause
•  Soft robotics in space

References

[1] Hauser, H.; Ijspeert, A.; Füchslin, R.; Pfeifer, R. & Maass, W. “Towards a theoretical foundation for morphological computation with compliant bodies” Biological Cybernetics, Springer Berlin / Heidelberg, 2011, 105, 355-370 http://www.springerlink.com/content/j236312507300638/

[2] Hauser, H.; Ijspeert, A.; Füchslin, R.; Pfeifer, R. & Maass, W. “The role of feedback in morphological computation with compliant bodies” Biological Cybernetics, Springer Berlin / Heidelberg, 2012, 106, 595-613 http://www.springerlink.com/content/d54t39qh28561271/

[3] Pfeifer, R. and Bongard J. “How the body shapes the way we think: a new view of intelligence” MIT press, 2006.

[4] Hauser, H and Füchslin. R.M. and Nakajima, K. “Morphological Computation – The Physical Body as Computational Resource” in e-book on “Opinions and Outlooks on Morphological Computation”, Chapter 20,
ISBN 978-3-033-04515-6, 2014, – http://tinyurl.com/pjvey43

[5] Nakajima K.; Li T.; Hauser H.; and Pfeifer R. “Exploiting short-term memory in soft body dynamics as a computational resource” Journal Royal Society Interface, 6 November, 2014, vol. 11, no. 100, 20140437 – DOI: 10.1098/ rsif.2014.0437 http://rsif.royalsocietypublishing.org/content/11/100/20140437.short

[6] Nakajima, Kohei, Helmut Hauser, Tao Li, and Rolf Pfeifer. “Information processing via physical soft body” Scientific Reports 5 (2015) Article number: 10487, doi:10.1038/srep10487 http://www.nature.com/srep/2015/150527/srep10487/full/srep10487.html

[7] Q. Zhao, K. Nakajima, H. Sumioka, H. Hauser, and R. Pfeifer “Spine Dynamics as a Computational Resource in Spine-Driven Quadruped Locomotion”  IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2013), 1445 – 1451, 2013

[8] Caluwaerts, Ken, et al. “Design and control of compliant tensegrity robots through simulation and hardware validation.” Journal of The Royal Society Interface 11.98 (2014): 20140520 http://rsif.royalsocietypublishing.org/content/11/98/20140520

[9] McEvoy, M. A., and N. Correll. “Materials that couple sensing, actuation, computation, and communication.” Science 347.6228 (2015): 1261689.

[10] “, F. Corucci, M. Calisti, H. Hauser, and C. Laschi “Novelty-based evolutionary design of underwater morphing robots” International Conference on Genetic and Evolutionary Computation (GECCO 2015) [accepted]

[11] Hermans, Michiel, et al. “Trainable hardware for dynamical computing using error backpropagation through physical media.” Nature Communications 6 (2015).

Baymax, Chris Atkeson and Hiro from Disney’s Big Hero 6. Source: Chris Atkeson

Baymax, the lovable, inflatable robot star of Disney’s recent hit, Big Hero 6, is far from a movie fantasy. With their soft cushiony bodies, robots like Baymax have very real prospects as future care-givers, space-travellers and more. Robohub’s Helmut Hauser spoke to the man who inspired Baymax – Chris Atkeson, Professor of Robotics at Carnegie Mellon University – about the hard science behind soft robotics.

[HH] Professor Atkeson, can you give us the background – how Disney approached you and how they decided to use a soft robot for the role of Baymax?

[CA] When Disney makes one of these animation movies, they do research. For example, for the movie Frozen, they had a bunch of animators who were from Southern California and had not really experienced snow. They sent a lot of them to Norway to play in the snow there and film it. That led to the movie Frozen.

For Big Hero 6, they knew they were going make a movie about a robot. They had the outline for the story from a Marvel comic, but they really wanted a robot that was new and different, they didn’t want a Terminator or a Transformer. So they contacted Disney Research, which tours robot laboratories around the world. One of the laboratories they visited was mine. They saw our work on inflatable robots to help dress, people, feed them and comb their hair, and the Disney folks were very happy because they’d found their robot: an inflatable robot that was very, very different from anything that had been in the movies before. And they also found a character for Baymax – a health or personal care, companion robot.

You are working in the field of soft robotics. Could you please explain what that is?

We’d like to have robots help take care of people and trying to do that with a very heavy robot is difficult, because sometimes computers break down or crash and the robot moves in unexpected ways, and sometimes people aren’t very co-operative. When I give a tour of my robotics lab to children, they often try to grab the robot by the legs and [it] could fall down. If a robot is very heavy, let’s say 100-150 kilograms, it can hurt the children or a pet. So I decided the best way to make a robot safe is to make it as light as possible, because mass is what causes gravitational force. Your momentum and your energy are proportional to mass. So, if we can reduce mass, we can reduce force and momentum; we can reduce energy and make it a much safer robot. So I was actually interested in light-weight robots – and maybe that’s what the field should be called – but we ended up deciding that the best way is to make inflatable robots and they’re inherently soft, so we got classified as “soft robotics.”

Can you also talk a little bit about the technological challenges that you have to face when you build inflatable robots?

Many people worry that an inflatable robot is too weak to do anything useful. So, one of the things we tested was: can they lift heavy loads? And yes, they can lift 10 or 15 kilograms and that’s enough for a lot of the care tasks that we want to do. If you don’t use a robot in the configuration of an arm of a humanoid and, instead, put the inflatable part underneath something, you can lift a larger mass. Inflatable devices are used to lift cars and buildings, so they can be very, very strong.

The second fear with inflatable robots is, if you puncture it, it will deflate and be useless. It turns out that the suits used for space walks, on the International Space Station, are also inflatable robots. They’re designed to self-seal if they’re damaged, so they don’t deflate immediately; there are techniques to help inflatable robots survive when they’re punctured, they can have compartments. So, puncturing an inflatable robot is not a big problem. In Big Hero 6, the robot repairs itself after it has several holes in it.

Yes, exactly – with scotch tape.

Yes.

Can you say a little bit more about energy efficiency? Is there a lot of energy needed in order to keep the pressure up in the robot or is that quite easy to do?

I don’t think it’s a lot of energy. It all depends on how much leakage you have. If you have a system that’s continually leaking and inflating, it may run up your energy bill. But if you have a situation where the structure doesn’t leak very much, and you only have to add a little air once in a while, that’s not a big problem.

Lightweight carbon fiber skeleton for an arm. Source: Chris Atkeson

You’ve already talked a little bit about possible applications for such types of robots. Do you have any other applications, outside of care-taking, in mind?

Robots that are toys and interact with children need to be very, very safe so I think soft, or lightweight, robots have an application there. Basically, [in] any kind of situation where you’re interacting with people, we need a high amount of safety.

Do you use any special sensors for your inflatable robots? Maybe a stretchable, bendable sensor, or do you use off-the-shelf sensors?

I think when you have a robot that’s not rigid, you tend to rely more on vision and external sensing. A rigid model can rely on sensors in the joints to achieve accuracy, but with the soft robots, you tend to have to look at where your hand is and where the target is in order to figure out: are you reaching the target accurately?

Closely connected to that, the control approach might be different as well, right? Or do you employ the same techniques that are used for a rigid robot?

I think the control approach may well be different. You can think of it as a much more flexible or complicated dynamic, essentially everything bends. And that means you have to be a little more cautious and use lower “control gains” or “stiffnesses” in order to move the robot around. So I think there need to be slightly different techniques, both in sensing and in control.

Your name is well-known in the field of machine learning too. I was wondering if you have any plans to link your expertise there to soft robots. Do you see the soft body as an advantage there?

We have already used learning for the “kinematics” of the robot; in other words, how to reach accurately, even though the parts of the robot are soft and bendy. The big surprise is that, when you start building complicated metal robots, hard robots, they also bend if they’re complicated enough – like a humanoid robot – because they have so many joints and there is play in them. So some of the techniques that we originally designed for soft robots are actually turning out to be useful for learning with the metal robots.

Baymax concept skeletons. Source: Walt Disney Animation Studios

In the movie, Hero builds rigid body parts for Baymax to enable it to fly around and to fight. Do you think that would make sense in real-life? And, if so, what applications could this have?

I think it’s a very good idea to have robots that can change modes. In this case, Baymax put on a hard outer shell in order to become stiff, and I think that makes a lot of sense. If you have a service or house robot, and it needs to be strong, it could use tools or put on a suit in the same way that people use tools. I think it’s a very cost effective way to get robots to do a wide range of things.

Do you also see a use for such types of robots outside the house, for disaster relief or for military use, for example? Because I would assume that a soft robot doesn’t have to model its environment in order to move around in it. Would that be a good application?

It depends on what kind of disaster you’re imagining. If you have an earthquake, for example, and a lot of rubble, you need small, agile robots to find victims, and very strong robots to dig them out. So maybe we’d have more specialized robots, say, in Japan preparing for an earthquake. If it’s a disaster that involves a lot of people getting sick, like the Ebola situation, then the soft robots may make more sense. They can be made very cheaply because we are using techniques similar to the ones used to make clothes, so we could even have disposable robots rather than trying to clean them. So, in the United States, when the Ebola problem was much greater a few months ago, there was some discussion about what kind of robots we should develop and soft robots was a possible answer.

What about using such types of robots for exo-planetary exploration, since, in space travel, weight and mass are a big issues?

Sure. Did you know we use inflatable landing systems to land probes on Mars? So, that’s already an example. If you’re going to build a habitat that people live in, it would probably be inflatable, if you want to deploy very large structures in space. Currently, we use unfolding wing-like things that are quite complicated. It’s possible that inflating them would be a better way to go – to make very large solar panels or antennas.

Baymax concept. Source: Walt Disney Animation Studios

You have a project called “Build Baymax”. Can you tell us a little bit more about that?

Yes. Of course, it’s inspired by the movie. There are two parts to the project. One is building the body: Can we build something with the capabilities of the robots in the movie? The second part is building the brain: Can we build robots that are capable of having the artificial intelligence that Baymax showed in the movie? We have groups to pursue those two targets, building the body and building the brain.

The movie is – I think – really inspiring, for adults as well as for kids. Because it talks about being a good scientist and being curious about things. So, if you were to give advice to kids who may be interested in soft robotics, what good sources are there for inspiration? What kind of skills does someone need to join the next generation of soft roboticists?

I would, of course, emphasize that studying mathematics and science is very important. But, I also think it’s possible for children to build soft robots, because it’s like making clothes and children can do it at home. If you want to build a metal robot, you typically need access to parts. With soft robots, it’s like building a kite and you can do that in your home. You can have some fun and get some experience. In the United States, we have a term, “home economics,” which is about teaching people to sew and things like that. Soft robotics could make home economics popular again.

Further reading on soft robotics:

Chris Atkeson’s webpage
Build Baymax website
Robosoft – An EU research community around soft robotics

• While most robots are either geek or jock, CHAPPiE is a lesson in sexy
  
• Robot love: Spike Jonze’s new sci-fi film ‘Her’ may be closer to reality than you think
  
• Automata: The new “sci-fi” blockbuster set to put robot ethics under a spotlight
  
• Ability to self-repair isn’t the same as consciousness: Analyzing Automata from a tech standpoint
  

See all the latest robotics news on Robohub, or sign up for our weekly newsletter.