Biomimetic Caterpillar Robots

On the biomimetic “soft robot” front, researchers at the University of Warsaw reported last month on an interesting design using liquid crystalline elastomer (LCE) that “wriggles” like a caterpillar when excited by laser light. The resulting wave of wriggles can move the small plastic sliver: it moves on its own muscle power. Different illumination gives it alternative “gaits”.

My understanding is that the light heats the material, which deforms. The “caterpillar” has layers which create travelling waves in response to the sweep of the laser.


The robot can execute various tasks such as walking up a slope, squeezing through a narrow slit, and pushing objects, demonstrating its ability to perform in challenging environments.

Anyone who has been invaded by caterpillars knows that their soft bodies and simple locomotion are extremely effective—they can get through the most unexpected entries.

This mechanism is entirely externally powered, the caterpillars have no power supply to run out. This particular version is steered by the external laser (and appears to have no ability to turn or reverse….). But I can imagine interesting things.

I could imagine a reconnaissance unit: a base station is delivered to a remote site that deploys and guides a swarm of “caterpillars”. The little pushers might deliver some sort of self-assembling origami robot, which might collect data and samples for return via caterpillar express. Rugged, light, and only the base station requires power supply.

  1. Mikołaj Rogóż, Hao Zeng, Chen Xuan, Diederik Sybolt Wiersma, and Piotr Wasylczyk, Light-Driven Soft Robot Mimics Caterpillar Locomotion in Natural Scale. Advanced Optical Materials (in prep), 2016.


Robot Wednesday

Tilting SeeSaw House

This month, performance artists Alex Schweder and Ward Shelley are presenting “ReActor”, which is a weird gimbolled “house” that tilts like a seesaw. (These artists are said to use architecture as a medium.)

The point is that the house responds to the presence of the two artists, tilting as they move, and, like a see-saw, tilting as the two of them move. This isn’t a practical design for living (as far as I can tell), it “is an experiment in how a house reflects—and even shapes—the relationship of its inhabitants”. When the two are in the house, they are acutely aware of the other’s movements at all times.


OK, this is certainly an impressive stunt. Not that this sort of gimbal is particularly ground breaking, except for the absurd scale of it. And people are aware of the other people in many houses by sound and other traces. If you live in an old house like mine, you know where everyone is by the creaking, shadows, and sometimes smell. (But if the floors start shifting, that’s very bad!)

There are also plenty of living spaces, such as dormitories, prisons, and some office buildings, where everyone is aware of everyone else. So this is scarcely a new concept, nor unexplored by normal people.

For that matter, today’s mobile devices let people track each other even when they aren’t in the same building.

So what’s the big deal?

The only novelty is really the modality by which they are aware of each other, which is fun-house crazy, and, frankly, very impractical. Anything with liquids will spill, stuff will fall off tables, and some people will be sickened by this motion. It’s basically idiotic as a real way to live.

As for the “experiment”, they seem to be excited by this “different modality of privacy, and different means of communications.” I don’t think they have found anything novel in the way people use this modality, at least not that I see.  In other words, it’s pretty much nothing as a social investigation, too.

  1. Diana Budds, This Experimental Home Tilts Like A See-Saw As You Walk Around Inside, in Co.Design. 2016.
  2. Omi International Arts Center. Alex Schweder + Ward Shelley: ReActor. 2016,

Wearable Thermoelectric Energy Harvesting

Melissa Hyland and colleagues at North Carolina State report some neat work with thermoelectric generators (TEG) in clothing. The idea is obvious: use otherwise wasted body heat to generate electricity to power wearable sensors and devices. With new flexible TEG technology, it is now possible to experiment.

The researchers point out that there are number of technical challenges. Generating electricity requires heat differential, as much as you can get. Waste heat off the body isn’t really all that significant, so it needs to be captured and routed to the generator. These collectors and devices have to be small, light, and comfortable to wear, and for practical clothing, they must be rugged (and washable).

Hyland’s group explored placement of their devices on several parts of the body, in a wristband and several locations on a tee shirt. Their results indicate that a patch taped to the upper arm generated the most power compared to the other locations, though, of course, a TEG in the chest of a teeshirt did generate some electric power.

They also found that airflow and the activity of the person affects the power output, generating more heat and heat differential.

The raw amounts of power are rather small, along the lines of 10 microwatts per square centimeter. This is hardly going to recharge you phone or anything like that. But it might power small sensors or other devices, such as medical or environmental monitors.

(I’m pretty sure that this is in the same ballpark as other wearable energy harvesting.  But I would need to check carefully to confirm this, so don’t quote me on this point.)

This technology joins other wearable power technologies, photovoltaic and kinetic power generation.  These techniques will be important, because we really can’t have our “smart clothing” needing external power or heavy battery packs.

Lightweight, Wearable Tech Efficiently Converts Body Heat to Electricity

  1. Melissa Hyland, Haywood Hunter, Jie Liu, Elena Veety, and Daryoosh Vashaee, Wearable thermoelectric generators for human body heat harvesting. Applied Energy, 182:518-524, 11/15/ 2016.

Book Review: “Are We Smart Enough to Know How Smart Animals Are?” By Frans De Waal

Are We Smart Enough to Know How Smart Animals Are? By Frans De Waal

Veteran animal psychologist Frans De Waal of the Living Links Center at Yerkes National Primate Research Center sums his life work in this marvelous book.

His career has spanned from the era of puritanical behaviorism to the current age of “cognition”.  During this period, the anthropocentric attitude that “evolution stopped at the human head” has been in retreat, as finding after finding have demonstrated that cognition, just like all other known biological systems, is continuous across species, and has evolved in every species for survival.

As he says, “Evolutionarily speaking, it would be a true miracle if we had the fancy cognition that believe we have, while our fellow animals have none of it.” (p. 43)  After all, we expect biological systems to be “continuous”, to operate on the same principles in all species, and to have similar functions in related species.

We don’t, he remarks,  study “rat livers” and “dog livers” and “human livers”, we study “livers”, and expect to learn general principles. Mental functions are the same.  “Cognition” in natural species has evolved, and must be investigated in that context.

One of the key implications of this is that cognition in any species has evolved for survival, has adaptive advantage, and does not develop “unnecessary” features.

Just as chimps don’t know how to swim, and whales know nothing of trees, the mind of a chimp and the mind of a whale have different abilities, relevant to the way they live.

Arguing about which species is “more intelligent” is kind of missing the point, and generally impossible. Yet this is what human supremecists do when they compare humans to other animals, hoping to show our “superiority” or ‘uniqueness”. Aside from the questionable assumption that people must be “special”, the evidence surely indicates that our cognition is fundamentally similar to apes, mammals, and animals in general.

In this book, De Waal marshalls evidence for complex mental processes in primates (his specialty), other mammals (elephants, dogs, whales), birds, and mollusks. Evidence is found in studies in the wild, and, De Wall emphasizes, in laboratory experiments as well. We need both forms of data.

One implication of the evolutionary perspective (as well as simple logic) is that experiments must be designed properly to be species-appropriate. If you want to compare humans and chimps, you have to be sure that the comparison is equally relevant (and fair) to both species. So many studies have found that animals do poorly one tests compared to humans are bogus because the task is human-oriented.

One telling case described by De Waal was a study that showed that domestic dogs followed human hand signals very well, but wolves did not (p.149). The initial conclusion was something like “dogs are smarter than wolves”, which is preposterous. A reevaluation indicated that wolves are highly intelligent, but simply do not follow directions from humans. They are very good at following directions form other wolves, though.

My own interpretation: wolves who are hunted by humans, and may occasionally hunt humans, do not trust humans. They trust other wolves. In this, I would say they are being very smart to not follow the instructions of a potentially deadly human, and they are certainly at least as “intelligent” as domestic dogs (who have “solved” the human problem by moving in and taming us).

Creating species-appropriate tests can be very difficult. De Waal recounts many clever experiments that successfully create tasks appropriate for the non-human participants. He argues that careful observation in natural settings or close to natural settings is critical for understanding the abilities and interests of animals, and for designing appropriate hypotheses and expriments. But, contrary to some ethologists, he sees that controlled experiments are necessary, too, in order to pin down the most plausible facts.

He makes the interesting “evolutionary” point about these two relationships to animals:

The study of animal behavior is among the oldest human endeavors. As hunter-gatherers, our ancestors needed intimate knowledge of flora and fauna, including the habits of their prey. Hunters exercise li==minimal control: they anticipate the moves of animals and are impress by their cunning if they escape…. A more practical knowledge became necessary when our ancestors took up agriculture and began to domesticate animals for food and muscle power. Animals became dependent on us and subservient to our wiil. Instead of anticipating their moves, we began to dictate them.” (p. 224)

I was particularly fascinated by the astonishing results about the complex social behaviors of apes and other animals. In particular, many animals will cooperate with each other, and some have very complicated politics with concomitant strategic ploys and plans. These quintessentially “human” traits turn out to be found in some form in many animals—but you have to know how to look. And, importantly, not every species or even individual is the same, it depends on what problems they have evolved to solve, and how they have grown up.

This is a wonderful book, inspiring and filled with fascinating research results. It is pleasing to see the intellectual victory of his “egalitarian” view point that respects every species, and soundly rejects human exceptionalism, which has been proven to be just plain false.  (You can tell where my own theoretical inclination has always lied.)

How Smart Are Animals? Any species that has survived thousands of years is, in fact, smart enough to make it. In some cases, this really smart indeed, though not smarter than “necessary”.

One more thing comes to mind. The overall thrust of this research shows that many animals have cognitive abilities that are similar or superior to humans, including cases of convergent evolution (where species have independently evolved similar mental functions). Human cognition, presumably including language and “consciousness” most likely evolved from ancestors with similar or at least “precursor” abilities. We don’t know how this evolution happened, nor do we have much idea what selection might have favored our cognitive abilities (though living in cooperative groups is certainly one such selective pressure).  It seems to me that this should be a rather urgent question for psychology, no?

Thinking further,  what do these findings tell us about machine intelligence? For more than half a century, humans have been whacking away at the problem, trying to create machines with a variety of human and superhuman mental powers. This quest has proved more difficult that originally hoped, and recent successes have often been in forms that mimic the function but not the mechanism of human abilities (such as probabilistic language recognition, and artificial “neural” nets). A ha! Convergent evolution!

But do we really know what we are doing? Probably not, because we don’t really understand the “components” of human (and animal) cognition, nor how they evolved.

For that matter, cognition clearly evolved in the context of evolving physical bodies, not in an isolated brain-in-a-jar. And for many cases, cognition evolved in social populations of organisms.

In this context, evolutionary approaches such as evolving robots and even genetic algorithms would seem to be on the right track. But even these cases are “evolving” isolated components, which isn’t likely to be the same as natural evolution.

But wouldn’t it be cool if we could figure out why different species evolved particular cognitive abilities (assuming we can even understand what “cognitive ability” means)? What if we had algorithms that, with the right parameters, generated faithful simulations of brains of chimps, humans, corvids, elephants, and whales? This would be a true model and theory of cognitive evolution, and it would let us create whatever “artificial intelligence” we wanted.

Phew! I wandered rather far afield! I think Sensei De Waal has inspired me to think deep and far. Can I say anything more complimentary?

  1. Frans De Waal, Are We Smart Enough to Know How Smart Animals Are?, New York, W. W. Norton & Company, 2016.


Sunday Book Reviews

Rosetta’s Last Hurrah

As most people know, the Rosetta spacecraft will be ending its mission to comet 67P/CG this month. If all goes as planned, Rosetta will brake and execute a slow dive, shooting back as much data as possible before impact.

Rosetta has been circling in, closer and closer to the surface, but on 29 September it will make its final monuevre (ESA prefers the British spelling), initiating a “free-fall slowly towards the comet” for about 20 km.

Rosetta’s last week at the comet. ESA
Rosetta’s last week at the comet. ESA
A simplified overview of Rosetta’s last week of manoeuvres at Comet 67P/Churyumov–Gerasimenko (comet rotation is not considered). After 24 September the spacecraft will leave the flyover orbits and transfer towards an initial point of a 16 x 23 km orbit that will be used to prepare for the final descent. The collision course manoeuvre will take place in the evening of 29 September, initiating the descent from an altitude of about 20 km. The impact is expected to occur at 10:40 GMT (±20 minutes) at the comet, which taking into account the 40 minute signal travel time between Rosetta and Earth on 30 September, means the confirmation would be expected at mission control at 11:20 GMT / 13:20 CEST (±20 minutes).

The project team has chosen to aim for one of the “active pits”, identified as a source of gas and dust. The final descent will pick up very close range observations of one such pit, to learn as much as possible.

This should be an exciting and fitting ending for Rosetta.


Space Saturday

Loomio Cooperative Decision Making Software

The Enspiral group (“more people working on stuff that matters”) is an interesting collection of cooperatives in NZ.  I would say they are “walking the walk” in a serious way.  their strategy is to boot up groups to tackle specific problems in an open, democratic and sustainable way. To do this, they have spun up an array of enterprises, addressing specific problems and needs.

One of the components of this constellation is the Loomio Cooperative dedicated to creating software to support decision making. As reported in their history, Loomio emerged from the Occupy movement, with the goal of making it easy to implement bottom up, democratic decision making, a la Occupy. There is quite a bit of information about how they are walking the walk, to be found online and in their handbook.

Specifically, the software is intended to solve the problem of “fast, inclusive, effective decision-making without meetings”.  “Without meetings” means asynchronous discussions, using digital technology. This has been done before, many times, so I was curious what might be new here, and why people are excited about it.

The project is open source (naturally), though the software is complex enough to make inspection a bit of a job. Not having time or resources to explore the source code, I looked at the documentation, which is pretty extensive.

The first thing to say is that the software is pretty simple, admirably so. It does only one thing, and it does that with a minimum of fuss. It is also a pretty portable and accessible tool, which I appreciate.  And kudos for multilingual support!

The basic point of interest is that Loomio implements a model of discussions and decision making that is clearly modeled on the Occupy movement. The voting mechanism even includes the “block” vote, so characteristic of Occupy meetings.

Of course, there is a lot of hard stuff that the software can’t solve. They describe a four step process:

  1. Gather (Invite the right people)
  2. Discuss (Have clear, on-topic conversations)
  3. Propose
  4. Decide together and Act

Well, sure. If we could do that, we wouldn’t need Loomio!

But seriously, there are important things that must happen outside the software. Finding the “right” people, is quite a loaded concept. (Who are the “wrong” people?) Having clear discussions is an art, and, as far as I know, using digital tools often is a formula for bad discussions. Software is not going to save you, only good people and good will can do this.

The proposal process is always hard work. A good discussion beforehand will surely help, but it is still necessary to get concrete, and to imagine practical and relevant actions. The software can’t give you good ideas, but it probably helps track the evolution of ideas.

Finally, there is the “decide and act” step. Obviously, software cannot create agreement where it does not exist. Loomio appears to work by making things clear about who agrees and disagrees, about what, and why. With good will, this may lead to consensus, or at least a well understood disagreement.

As far as moving to action, obviously software isn’t really going to make that happen. But it does provide open documentation for the proposed action, which should be helpful.

My point isn’t to pick at Loomio. I’m just making clear that this is decision support software. Loomio does not make things happen, only people can do that.

What then is “special” about Loomio? What makes its fans happy?

To a first approximation, it does the same thing as lots of other software. In fact, you could do everything with email, if you were willing to do some work to archive things carefully.  Some of the work Loomio does definitely helps: for example, Loomio enforces deadlines, and automatically archives everything, so there is a transparent record.

From what I have seen, though, the best thing about Loomio is the instructions. The software is pretty standard, but their instructions, case studies, and “how tos”, model the kind of democratic decision making they believe in. I think you could probably do pretty well following Loomio’s instructions without the software.

I suspect that Loomio also does the most quintessential “magical” thing a software tool can do: it makes it easy and fast to do things “the right way”. Loomio has strong ideas about what is the “right way” to collaborate, and they make this as easy as possible. (I bet you can “misuse” Loomio to have bad discussions, non-democratic decisions, and generally disenfranchise people—but you would be working against the grain. Why bother?)

Is Loomio the be all and end all solution? Of course not. But if you are doing distributed collaboration, especially about something in the digital realm, then you need this kind of software. And Loomio might be good for you.

  1. Cat Johnson (2016) The Loomio Handbook: A Roadmap for Worker-Owned Cooperatives. Sharable,
  2. The Loomio Cooperative, The Loomio Cooperative Handbook. The Loomio Cooperative, Aotearoa New Zealand, 2016.


Blockchain Technology for Financial Data?

This week there were several press releases about “private” blockchain technology, which are thin pickings information-wise, but suggestive of the state of play. E.g., this and this but also this.

Basically, many companies, especially in financial businesses, are looking at blockchain technology to simplify (and vastly cheapen) their “back room” operations. (Say goodbye to thousands and thousand of jobs!)

For this use case, they don’t want a public ledger a la Bitcoin, they want the same thing but on a private network. This variation begs the question as to what parts of the Nakamoto project do they need, and how are they going to use it? Furthermore, given the massive amounts of IT they already have, what is the “special sauce” provided by the blockchain, per se?

These are proprietary efforts, so there are limits to what I know. But let’s see what we can infer from one of these press releases.

R3 and Axoni announced a successful (in their own view) demonstration of how ot use a private blockchain to manage financial data, specifically records surrounding bond offers. The demonstration illustrated how multiple parties can have a “real time” trusted access to information about “which parties on the ledger have created, issued and proposed amendments to the data record”.

Judging from other similar projects, the actual documents and data exist in other systems, e.g., “curated” by one party or another. The blockchain seems to carry a cryptocgraphically signed version history of these documents. The idea is to be able to know, instantly and reliably, which is the current version of the information in question.

Assuming I have the idea basically right, then the system is making use of public key cryptography in an important way, though this isn’t “blockchain” technology, strictly speaking. The blockchain appears to serve as a reliable write-once store, and the consensus mechanism gives everyone confidence that everyone has the same information. It can still be wrong or dishonest, but everyone has the same bad stuff.

(How “real time” this might be depends a lot on what you mean by “real time”.)

I don’t see that they need the currency and mining protocols at all in this usage. The private network does the consensus protocol as part of the cost of participation, and there isn’t really any point to the alleged “incentives” of the Nakamoto mining process.

So how are they actually using the blockchain? There must be protocols similar to version control that post updates, properly sealed and signed. This presumably uses digital signatures to effectively notarize the documents and data objects, certifying who did what, when. The actual documents are not posted, but cryptographic checksums assure that anyone can confirm if a particular copy is valid or not.

The blockchain serves as a write-once bulletin board for these notarizations that cannot be tampered with. The consensus protocol essentially creates time stamps without trusting any single party or system. The time ordering is very important for accurately tracking changes and determining the “current version” of information.

The most important point, though, is that the systems must incorporate processes for efficiently pushing out information, properly sealed and signed, and for reading off the blockchain, properly checking seals and signatures.

These processes must, at their heart, have rules for who can sign what, and mechanisms for confirming the authority of the public signatures. Everything depends on managing secret crypto keys, and on having a mechanism for establishing which parties are trusted to do what.

These are challenging problems, but surely possible within a private network. They are not, however, specifically “blockchain” problems, nor does a blockchain help solve them.

So–is a blockchain necessary to build this system? Probably not. There are other ways to create distributed databases on a private network, and they probably perform similarly.

I would guess that a blockchain might be relatively inexpensive compared to other approaches, though given all the other components (they use a secure private network, for goodness sake!) who knows if that is significant.

Perhaps the distributed, Nakamoto consensus based timestamp mechanism is the most important feature. The fully distributed, non-disputable ordering of the records is a unique feature of Nakamoto style blockchains, and it is very useful.

In other words, a blockchain based ledger may actually be a very useful way to do this kind of version control across multiple independent parties.

But you have to implement the rest of the picture, including the human protocols to reliably use the system.  The blockchain is only a tiny piece of the needed system.


Cryptocurrency Thursday






A personal blog.

%d bloggers like this: