Android Robotics up to 2019: The real story; in 5 parts; part 1


    Quite a long time ago, seven years ago to be precise, i wrote a series of posts describing the state of android robotics in the world. At the time i was a high school student, with a keen interest in android robotics, who absorbed a bit of knowledge from English, Japanese, Chinese, Korean and Russian internetz and wanted to spill it somewhere.

    While the posts were not too professional, and not to my standards of today, they were worthy enough to get stolen and even get translated by unapproved English Habrahabr mirrors, and to this day, appear in searches.

    After those posts were written, Habrahabr got split. Removal of everyone outside of pure coding who were considered «not cake enough» to Geektimes felt like an insult and so i left the platform. Yet, the website was reunited last year, and much to a personal surprise, fairly recently an English version of Habrahabr was released.

    During all these years i managed to be kicked from one university, finished another with a thick thesis on «Usage of Robotics in Disaster Conditions», lived in the Republic of Korea for half a year, and most importantly, not only expanded my knowledge of android robotics in such ways that the Robotics folder on the main hard drive is now more than 300GB in size, but also expanded the knowledge via journeying and personally meeting projects of the past and present, creating quite a decent archive on Youtube and met not only with the robots, but the engineers and scientists as well.

    While i am still nowhere to be a robotics engineer, (and in the daily life i attempt to be a traditional slice-of-life artist), i feel that my tiny gigabytes of knowledge might be worthy of sharing, and today on Habr i'm publishing the real story of Android Robotics from the beginning up to 2019.

    Forgotten in the previous decade, androids are in the hype now. Detroit: Become Huaman (oh, don't let me started on this mess), Westworld (the first season is actually decent, the second one i have not seen), news about sex robots and sex robot brothels (those will definitely be mentioned), Sophia saying something weird on some publicity stunt (that too will be mentioned), this is all mainstream news now.

    Yet, with all of this coming to our screens, even the Wikipedia article about androids is lacking. There are no dedicated android robotics websites, not much in terms of books, scientific articles in the field are not attracting much attention and are hidden beneath the paywalls, the engineers and scientists are also usually hidden and don't have much interest in spreading knowledge which gives way for hype-seekers and publicity-stunt lovers, even the random news about the field is spread between English, Japanese, Chinese, Korean, Russian and German sources without much translation in between.

    The reason for writing such a story is to attempt to evolve this chaos a little bit, spread the personal knowledge about the field and to create a sort of a public lecture about android robotics in a text form that can be indexed and referred upon.

    The story will mention different robots, but focus will be on android robots (beginning with specifying what the term even means). This will not be a list of all androids in all the history, and there definitely will be intentional omissions (lack of numbers such as degrees of freedom), unintentional gaps, and mistakes. This will be a long read, but hopefully, a fun and informative one.

    Before i begin, however, i'd like to to ask everyone who would like to message me about language-related things (#SecondLanguage will be strong with this story) such as grammar and typo corrections to do so in private messages, otherwise the comments would just be filled with only comments on grammar and typos.

    The first draft of this story featured all the illustrations in the form of embedded videos, but with more than a hundred videos present, the preview page just crashed. The next attempt was to replace all embedded videos with .gif images, but unoptimised .gif files made the page 1.4GB in size and those could only be shrunk to 300MB.

    Videos i find necessary, as it is always important to see robots in action. To fix the issue and still retain videos, again the embedded Youtube videos, the story is split in multiple parts, with Youtube videos illustrating most of the robots, with a few exceptions. There are no other reasons for splitting, so treat the series as one big text.

    Videos by the user «Vokabre» with titles present are my recordings, but those without a name are re-uploads with the original footage either deleted from Youtube, or sourced from elsewhere, mostly from P.R. China. Where possible, i will give proper attribution, but unfortunately, it is often impossible to find a source of the original footage, especially in the internet of P.R. China.

    Another thing to note, just not to repeat it down below all the time, every time the story mentions «currently», «at the moment», it means January 2019, and it is from this point all the «a year ago» or «recently» remarks should be counted as well.

    «If language be not in accordance with the truth of things, affairs cannot be carried on to success»
    Confucius, Analects, Book XIII, Chapter 3, translated by James Legge
    Among geeky public it is a fairly common knowledge that the term «robot» first appeared in 1920s play «R.U.R.» written by a Czech author Karel Čapek. The word, derived from the general Slavic root for «work», and Czech word for slave work, became attached to electromechanical machines performing some king of physical task in the physical word, and machines of human form in particular.

    The term «robotics» was also invented by a writer, this time Isaac Asimov in the late 1940s. The term is now used to describe the science and engineering field related to robots.

    In Russian, the term slightly differs from the one in books, translations would usually say «robotehnika» (роботехника) while the real-world term is «robototehnika» (робототехника).

    «Robospychology» invented by Isaac Asimov recently had arrived to the real world too, but with a differnent name and in a slightly different form of «roboethics».

    The definition of a robot is vague. There can be numerous definitions, and books on introduction to robotics would often have a few pages saying just how how vague the definition is. Masahiro Mori, (to whom we will return later), famously said that defining a robot is a hard as defining a Mount Fuji. You can see the peak, but you can't see where the mountain begins.

    For example, is it fair to call a remote-controlled vehicle a robot? In disaster robotics this is precisely what happens. Should a machine with a pre-recorded routine be called a robot? Modern animatronics is as complex as many robots, but the units run a pre-recorded routine. Can a virtual agent be called a robot? Quite often we see this happening too.

    For not to be too nitpicky, for this story we'll just let everything that's being called «a robot» to be a robot, which would in all cases mean an electromechanical entity capable of movement in a physical world.

    «Artificial intellegence» is a complex topic, and this story would barely touch it. While it might not be too scientific, here, all the robots are briefly divided into having no autonomous function (teleoperation), autonomous performance of a pre-recorded routine (or an idle motion), pseduo-AI (an imitation of artificial intellegence with a complex set of if-then with none or barely any learning capabilities), weak AI (true artificial intellegence in a modern sense of the term, this what is usually referred to when mentioning «artificial intellegence research»). Advanced general intellegence (AGI) barely makes an appearance, being a theoretical concept and rarely appearing as part as android projects.

    The term «android» actually predates a «robot», it appeared in middle-19th century as one of the terms to describe an «automaton» (a mechanical robot so to say, more on this later). The term is a Greek neologism meaning «having a likeness to a man».

    Later, a term «gynoid» was invented to mean «having a likeness to a woman», but although this term was popular in old science fiction and is somewhat popular among love doll enthusiasts, usually woman-shaped robots too are described as «androids» which is seen as gender-neutral for most, and the word «gynoid» remains obscure.

    George Lucas invented a term «droid» based on the term «android» to describe robots in Star Wars, but in the real world the term wasn't accepted to mean anything in particular.

    The term «android» describes a robot with a full human-like appearance (e.g. Terminator), while the term «humanoid robot» describes a robot with a human-like shape (e.g. C3PO). Of course, in science fiction and even in the real world the terms often mix up, and even Google managed to call their OS «Android» while using a humanoid robot as a mascot.

    The term «cyborg» is often thrown in the mix, such as in 90s translations of Terminator to Russian, Terminator was called «cyborg» yet in reality the term «cyborg» applies to either humans with mechanical enhancements, or more often to theoretical robots utilising biological parts (e.g. Robocop). While occasionally called «cyborgs», «replicants» from Blade Runner are hard to classify, as those are basically modified humans, and thus fall into a separate category.

    The words «android», as well as «robot», were loaned to multiple languages including Russian (андроид, робот), Japanese (アンドロイド, ロボット) and Korean (안드로이드, 로봇). In Russian, the term «anthropomorphic robot» (антропоморфный робот) is also occasionally used in tech and science to describe an android.

    For a short while, in roughly 2010-2015, a term «avatar» (аватар) was popularised in Russian media as a term for androids, alluding to a remote control nature of some androids (such as in the film Surrogates).

    In English, in news articles and in the web in general is quite often to see descriptions such as «human-like robot», «lifelike robot», «realistic robot» being used to synonymously to the term «android».
    In Chinese (where the term for a robot is jīqìrén, 機器人 «machine human») an android can be called «simulation robot» (仿真機器人, fǎngzhēn jīqìrén) and an android of likeness to a woman is often described as a «beauty robot» (美女機器人, měinǚ jīqìrén).

    Now, let's return to the automatons or automata.

    The terms describing «automatons» in multiple European languages are based on the same «automat» root, with «automaton» being the most common English and Russian (автоматон) way to describe a mechanical doll (although just a phrase «mechanical doll» is used too).

    Later automatons can be electro-mechanical but still be called automatons and not robots. The general rule would be that in electro-mechanical automatons the motor is used to power the network of gears (with this network originally being powered by water, spring or even human power), and there are no more motors or other fancy things such as sensors and electronics beyond this initial push.

    Historical roots of automatons are vague, as mechanical figures capable of automatic movement existed since ancient times, with water-powered figures in ancient Greece being a common example from history books and dictionary articles, but automatons appeared not only in European and Western culture in general, they existed in China, in Korea and in Japan. Note that since the earliest examples it is usually humans who were represented in mechanical form. The «Moores» that strike the bell on top of Saint Mark Clocktower in Venice are a good example of a pre-Modern automaton.

    Golden age of automaton-making in Europe begun in the Early Modern period. From 17th century to early 20th century, automatons became the pinnacle of watchmaking. Figures capable of realistic movement were expensive curiosities bought first by the wealthiest of the wealthy, and later, as automatons became simpler to make, by just the wealthy people.

    There were even philosophical implications to automaton-making, as at the time the world was generally perceived in terms of classical physics, thus, the argument was that with having cogs tiny enough and the machine complex enough one can recreate even a human being, who is, in fact, itself can quite be an automaton.

    (This paved the way to such hoaxes as "The Turk", a chess-playing automaton who turned out to be a doll remotely controlled by a person cleverly hidden inside the base. We now understand that chess was hard to crack using sophisticated programming, but at the time many believed that a complex network of cogs was enough).

    Jacques de Vaucanson from France, who is also known as an inventor of an automatic loom utilising punched cards (first time in history when punched cards were used for «programming») created a few automatons in his days, most notably he in 1739 unveiled a «Digesting Duck» automaton. The original automaton was lost, but there exist recreations such as this one kept at the private collection of oligarch David Yakobashvili (founder of Vim-Bill-Dann). The video is sourced from the link above.

    Birds were a common source of inspiration for automatons. After John Joseph Merlin and James Cox created the «Cox's timepiece» in Britain, in 1773 they two built the «Silver Swan» automaton. It is definitely one of the most beautiful automatons of the era, and it remains in working order to this day. (The Swan is currently exhibited at the Bowes Museum in England and performs for the public every afternoon at 14-00).

    To those who are like me are unlucky ones who won't be granted a visa to UK, another of James Cox creations can be seen at the Hermitage Museum in Saint Petersburg. The Peacock Clock was created in 1772, then shipped to Russia, and then re-assembled by Ivan Kulibin. Quite a feat considering there were no manuals and no one to consult with. Peacock still works, and while Hermitage is surprisingly secretive about the times of the Peacock's performance, it seems the current times is each Wednesday on 20-00. On a few occasions i've heard boasting that the Peacock contains most of the original details compared to all the automatons of the era, but i doubt that anyone properly conducted a research into this.

    All of the birds got me distracted from the main topic, but with the next name we're properly returning to the land of human-like machines. The name of Jaquet Droz would be recognised by horology enthusiasts, it is a company that produces elite watches to this day, but originally it was the family name of two watchmakers from Switzerland.

    From 1768 and 1774, Pierre and his son Henri Louis, two of Jaquet Droz family, with Jean-Frédéric Leschot, Henri Maillardet and Jacob Frisard constructed three automatons that can be cited as the pinnacle of the mechanical enginnering of the golden age of Automata.

    (There also supposed to be the fourth automata, «The Grotto» (La Grotte), but it was lost and likely destroyed during the French Revolution. The automata is described as being a huge landscape scene with a flute player playing to a shepherdess).

    The Draughtsman (Le dessinateur) is an automaton of a young boy who draws one of three pictures stored as a collection of x and y analog coordinates hardcoded on a spinning cog, a pictures of a cupid, of a dog, of a royal couple, and of a Louis XV. In addition to the hand, the boy's eyes and head is animated, and during writing he stops to blow the graphite dust from the table.

    The Musician (La musicienne) is an automaton of a lady who plays a few melodies on a specifically constructed organ. In addition to the Draughtsman's animation, there is gentle torso movements and breathing action. The music is again hardcoded, but this time in a curious manner, there are different codes for different tunes, and in between them there's an idle animation which either plays before beginning a next tune or plays as an idle movement after playback of the full playlist is finished.

    The Writer (L'écrivain) made out of triple the pieces of the previous two automata, and is the most complex of all three. Similarly to the Draughtsman, The Writer is also a young boy, but this time he writes, and his text is not hardcoded. Instead, he is programmable to write up to 40 characters (properly spacing them, and dipping the feather in ink when needed). To this day, all three automata perform at the museum of Art and History in Neuchâtel in Switzerland, and the Writer automaton is occasionally being reprogrammed to write something new for different events.

    While these three toured the world and became quite famous back in the day, they had also shown the limit of technology. Future automatons were simpler and with no similar programming as the Writer had. I've read mentions on how different atmospheric conditions, and differences in temperatures affecting the expansion and contraction of metal cogs changed the behaviour of the automatons — not very practical when the main task of the machine is to impress people and suddenly something breaks because it's too hot or cold.

    Still, in nineteenth century Swiss and French automaton makers created hundreds of doll-size automatons, which to this day are often remain in working condition but unfortunately, mostly in private collection and not in the museums.

    As mentioned before, some of the later, early 20th century automatons utilised electricity to power the initial movement, but with all the social changes in 20th century, and numerous wars, the market for automatons disappeared for a while, and the advantage of mechatronics halted automaton-making even more.

    Still, like as many watchmakers continued making mechanical marvels, some automaton-makers continued making new automatons, either fully mechanical, or powered by an electric motor. For example, i am personally quite fond of automatons by Franz Oehrlein who lived from 1932 to 2013. These automatons go for high prices and are either commissioned by or sold to collectors, but fortunately some end up in open collections such as this prototype exhibited at Kawaguchiko Music Forest in Japan.

    Some years ago Moscow Polytechnic museum had obtained an electro-mechanical doll to represent automatons, made by a Saint Petersburg maker Alexander Getsoy. While somewhat crude in movement, and slightly broken while at the museum (i wrote both to the museum and to the maker to no avail, the museum doesn't care enough), the doll still operates at their «Russia Makes By Ourselves And Whoever Disagree Will Be Nuked» exhibition at the All-Russia Exhibition Centre in Moscow.

    At the same time as Europe had their early modern period, Japan had another period of growth and prosperity, and unlike how it was in Europe, a period of unity and peace. The civil war of before had ended, and Japan was unified under Tokugawa military government, with the headquarter in Edo (modern-day Tokyo). As early as in seventeenth century, Edo became a megacity with a million inhabitants. The growth in wealth resulted in the growth in literacy, and Edo of the era featured some of the things that weren't even present in Moscow up until recently such as food delivery.

    Printmaking was booming, art in forms of prints was distributed for a price of a bowl of soba noodles, with souvenir prints of famous places, portraits of famous women and men, and a few erotic prints. Theatre was booming too, and while Edo is mostly associated with the growth of Kabuki, Bunraku puppet theatre was quite popular as well, with puppets not seen as «just for kids» like in Europe. At the time Japan was closed from the outside world, but there was trade and most importantly trade in knowledge with Joseon (Korea) and via Dutch East Indian company at the port of Dejima in Nagasaki.

    European knowledge of clockmaking was imported as well, and local masters had created clocks suitable for usage in Japan, where traditionally hours differed from the time of day and time of the year. I myself had seen some of those clocks, even alarm clocks, and then watches, and it took me a while to even get basic understanding of the mechanics involved. As in Europe, Japanese engineers had dived into mechanical puppets as well.

    The new art was born, called the art of Karakuri Dolls (からくり人形, karakuri ningyō). In theatre the dolls were usually controlled by a puppeteer who was in charge of operating the doll using strings, but there also were small dolls similar to European automatons which were shown as curiosities, and giant dolls which were placed on floats and displayed on festivals. Similarly to this, Korean and Chinese mechanical puppets created at the same time were utilised in giant clocks. As in Europe, in the 20th century the art of karakuri ningyo was halted for a while, but it returned, and for the World Expo 2005 in Aichi, (which will be mentioned later again) a new mechanical doll similar to the ones used on the floats before was created, which is still performs at Expo 2005 museum near Nagoya.

    One name often associated with karakuri dolls would be of Hisashige Tanaka, a great man born in Kurume in modern day Fukuoka in 1799, who not only built karakuri dolls, but also studied so much of European knowledge, that he in Saga later built the first Japanese steam engines, including one used on a steam boad, and one used for a model of a steam ship. After Tanaka died in a new westernising Meiji-era Japan in 1881, his son continued the father's research, founding Tanaka Engineering which later was renamed to Shibaura Engineering and after merging with Tokyo Electricity (Tokyo Denki) became Tokyo Shibaura Denki or simply Toshiba.

    Twentieth century was celebrated by having two great wars, quite a few civil wars and a few genocides, and only after WWII things settled enough so developments on robotics would become possible. As mentioned before, the concept of robot appeared in late 1940s and people got quite fascinated by an idea of a mechanical human, with robots being as popular in science fiction as space exploration.

    Most robots of the time were imagined as being humanoid, and android robots were explored in science fiction as early as in 1949 when the first part of «Elijah Baley and R. Daneel Olivaw» sci-fi detective stories by Isaac Asimov was published. R. Daneel Olivaw is an android by all the modern definitions, a human-like electro-mechanical robot made out of mechanical parts, with the only «fiction» part being, like with all Asimov's robots, the positronic brain. The series even went into such subjects as roboethics and robosexuality half a century before the discussion on this topics were to reappear in the real world.

    The advantage in mechatronics, and economical growth in the US, West Germany and Japan paved the way for the dream to became reality. Industrial robots were patented in 1954 and after a decade or so arrived to the factories to do actual tasks. Research labs diving into fresh new computer science field often continued their research into not only artificial intellegence, but robotics as well. Humanoid robots of the time were mostly crude novelties, and proper research happened on wheel-based robots, so androids remained in science fiction.

    There were some examples of early electromechanical robots which could pass a definition of an android such as «She» from 1934, but the breakthrough came later, and came from Disney. Disney in 1960s had invented what they called «audio-animatronic», a mechanical puppet that performs pre-recorded movements synced to a pre-recorded audio. For the 1964 World's Fair in New York Disney had built four attractions, all of which utilised the new audio-animatronic technology.

    From simplest to complex the attractions would be «Magic Skyway» commissioned by Ford, «It's a Small World» commissioned by Pepsi, «Great Moments with Mr. Lincoln» commissioned by the State of Illinois, and the «Carousel of Progress» commissioned by General Electric as part of «General Electric's Progressland». The last two featured complex audio-animatronic humanlike figures that by all definition could be called androids.

    Disney utilised audio-animatronics in their films, beginning with Mary Poppins that released by the same year, but it is the attractions where the audio-animatronics did shine the most. Even Isaac Asimov himself visited the World's Fair and imagined the World's Fair of 2014. His description was actually not too far of from reality.
    Robots will neither be common nor very good in 2014, but they will be in existence. The I.B.M. exhibit at the present fair has no robots but it is dedicated to computers, which are shown in all their amazing complexity, notably in the task of translating Russian into English. If machines are that smart today, what may not be in the works 50 years hence? It will be such computers, much miniaturized, that will serve as the «brains» of robots. In fact, the I.B.M. building at the 2014 World's Fair may have, as one of its prime exhibits, a robot housemaid*large, clumsy, slow- moving but capable of general picking-up, arranging, cleaning and manipulation of various appliances. It will undoubtedly amuse the fairgoers to scatter debris over the floor in order to see the robot lumberingly remove it and classify it into «throw away» and «set aside.» (Robots for gardening work will also have made their appearance.)

    New York Times, August 16, 1964, Visit to the World's Fair of 2014, by Isaac Asimov

    Experimentation with animatronics and robotics became common in films for creating realistic puppets for practical effects. The attractions from the World's Fair have made it into the Disney Parks, and from now on it was a norm to have audio-animatronics or how the rest of the English-speaking world called them, animatronics for the dark rides. Many of the animatronics were human figures, and it is the entertainment field where first of de-facto androids had arisen in the modern times, and this field will to this day be one of the major applications of androids.

    Cybot Corporation from the 1980s Tokyo might be the first business that specialised on development and construction of androids. The company founded by Mizuno Shunichi had created androids for store displays and exhibitions, at least one android was made for a sex museum at the popular at the time Atami resort not far from Tokyo. Their androids were mentioned in news articles beginning from 1982, and there exists a news recording in French showcasing most of their robots at the time, including their most famous Marilyn Monroe robot. Like with Abraham Lincoln by Disney before, androids are often utilised as a way to reconstruct a historical figure.

    Mizuno Shunichi later had moved to Waseda University known for their early developments of advanced humanoid robots in the next decade. The move might have been caused by an economic crisis that heavily affected Japan in the 90s, and out of which only recently Japan had managed to climb.

    With entertainment industry slowed down, it is academia that continued the research into androids. Robodex exhibition in 2000 among other new robots featured Saya by Hiroshi Kobayashi Laboratory of Tokyo University of Science. The laboratory that had opened in 1999 was researching into different robotics projects, with exoskeletons and androids being two major focuses for the first decade.

    Saya was a simple android, with only the face animated, but unlike animatronics of the past Saya was not just used for show, Saya was a tool for human-robot interaction research. Saya was also experimented to perform tasks as receptionist robot. The last experiment was a proof-of-concept of using Saya as a teacher.

    Over the years, multiple versions of Saya were constructed, with at least two shipped overseas, one to Tel Aviv University and one to ETH Zurich. The project lasted for almost a decate and was put aside in around 2010 as Saya was not quite up to the standards of androids at the time, and most of the research that could be done with her was finished. Tokyo University of Science appeared later, in 2007, at an exhibition with a robot Pikarin, but nothing much was made public about the project, apart from a few photos no mentions exist online with descriptions that the robot was made in the likeness of a student named Hikari Asano.

    Early 2000s was the time when the most prominent face in the android robotics of today had emerged — Professor Hiroshi Ishiguro of Osaka University. The quiet man begun researching androids and human-robot interactions with androids with building a robot copy of his then-five-year-old daughter. The robot was named Repliee R1.

    Since the following video is a part of a Japanese documentary that shows Repliee R1 and mentions the famous graph for the «Uncanny Valley», it is a good place for an intermission to explore the concept. «Uncanny Valley» is a theory proposed by Masahiro Mori who was mentioned before for his comments on the definition of the term «robot». Masahiro Mori born in 1927 is one of the first roboticists, a founder of one of the most prominent robotics competitions Robocon, and he was mentioned before in this story with his comments on defining a «robot»

    Masahiro Mori proposed this idea back in 1970s that with the growth of human-likeness we, humans, grow sympathy towards an object, but after a certain level we became disgusted or even fear an object, dropping into a «valley» on the graph of human response to human-likeness, and only after this valley is «crossed» with the object being truly human-like we again return to be attached to the object. The fall is more dramatic when the object is moving, and less when it's static.

    Not only robotics was affected by this concept. Sculptures, dolls, paintings, puppets in theatre and in animation, there are quite a lot of static and moving sources for causing the effect. Notably, until quite recently CG animation had troubles with recreating human figures, with uncanny valley effect present in titles such as Final Fantasy: The Spirits Within from 2001 and The Polar Express from 2004.

    Even more often this happens with games, and titles such as BioShock from 2007, Heavy Rain from 2010, L.A. Noire from 2011 are often mentioned as riding their players straight into the uncanny valley. While today's CGI can go beyond the uncanny valley, a simple mishap in facial capture could easily trigger the feel, and CG animation studios often choose a more «cartoony» look to keep staying «before» the valley.

    As a research from 2011 shows, the effect might be caused by a dissonance from the image of how a human should look and behave that is pre-written in our brains, to the data that we gather by observing.
    Published in the Oxford University Press journal Social Cognitive and Affective Neuroscience, the functional MRI study suggests that what may be going on is due to a perceptual mismatch between appearance and motion.

    Your Brain on Androids, July 14, 2011, By Inga Kiderra

    uncanny valley might as well be an ancient mechanism to trigger a disgust response to people with mental and/or physical illnesses, and this mechanism might be the one that still causes trouble in our perception of others to this day.

    Repliee R1 was built utilising electric (servo) motors, but for the next project Hiroshi Ishiguro collaborated with Kokoro Dreams, an animatronic company which previously made dinosaurs for attractions. The collaboration was mutual, in January 2004 Kokoro Dreams produced Repliee Q1 robot and before that in November of 2003 the sister robot Actroid, with first remaining at Osaka University for research, and the second remaining at Kokoro Dreams as an entertainment robot to attract guest eyes' at exhibitions. On the first video is Repliee Q1, on the second is Actroid.

    Both robots utilised air (pneumatics) instead of electric drives, as it was proven by the entertainment industry to be a more efficient way to recreate smooth movements. The cons are that a pneumatic driven robot requires a compressor, which can be quite noisy, and an air powered robots are difficult to make mobile. With andoids unlike most humanoid robots sitting or standing at one place the latter was not in issue.

    Repliee Q1 and the next androids Repliee-Q1-expo (constructed in December 2004 in time for Expo 2005), and Repliee Q2 (constructed in July 2005) were a platform for experimenting with uncanny valley avoidance. It was noticed by Hiroshi Ishiguro's team that idle animations and unconscious movements like blinking are as important for creating an illusion of human presence. Multi-directional microphones and cameras were experimented to help software understand the locations of people around. Concepts such as programming negative response to unasked touching were tried as a way to make behaviour more realistic. One of the usual demo programs for Repliee Q1-expo and Repliee Q2 involved interviewing a guest, but this was a pre-recorded routine.

    Note, that there is a bit of a confusion between these three robots, with Q1, Q1-expo and Q2 being used interchangeably and attached to other robots as well. I myself am not even sure if Q2 is a separate robot or is just an updated version of Q1-expo. There was also talks about Q2-bis that was never was released, which was planned as a second version of Q2, either as an update or as a new robot.

    Kokoro Dreams kept developing the Actroid series independently (or rather, with less involvement of Osaka University). For the upcoming World Expo 2005 in Aichi east of Nagoya in March 2005, Kokoro prepared four Actroid robots. Three of them (occasionally called «Actroid Expo») were sitting in information booths at the east, west, and north gates, and the fourth one was a standing model performing at the stage. The routine for the performer was pre-programmed, but three others worked autonomously.

    All the robots were created in conjunction with NEDO, «New Energy and Industrial Technology Development Organization», an independent administrative institution of Japan (JAXA, Japan Aerospace Exploration Agency, has the same status). The robots were air-driven, and were quite impressive, especially the three information booth androids which, while running an if-else routine when speaking, and having a limited number of questions, managed to perform voice recognition in three languages, Japanese, English and Chinese in a crowded environment, and do this quite well.

    Like Saya, these robots had explored another field where androids can shine, the service jobs. Info screens and virtual characters can decently replace human interaction in many scenarios, there is still a big demand for human-like interactions, and we see this with the growth of voice assistants (even if we, being geeky, might personally avoid those preferring a good old keyboard) and robots such as Pepper.

    It is at that time, when Hiroshi Ishiguro coined a term «android science», describing a new interdisciplinary field emerging when robotics and artificial intellegence studies clash with humanities, sociology, psychology and even art.

    After the exhibition had ended, the stage and the north gate Actroids were relocated to the Memorial Museum of Expo 2005 where they stay static (although i have seen mentions that they were operating at the museum too for the first few years after the relocation).

    One of the Actroids, the girl named Sakura from the west gate moved to the older Tokyo Science Museum that has a designated NEDO area, and to this day answers the questions of the visitors. The dialogue was changed to be about the museum, only Japanese language was left available, but the routine remains the same.

    The dialogue is constructed in a clever manner, it resets after three exchanges, eliminating long waiting lines and solving a continuous problem for all the imitation AI and weak chatbot AI systems of the lack of long-term memory and difficulty of constructing a long dialogue. For those who can read Japanese better than me, NEDO published an article describing the software and the voice recognition.

    After the Expo, Kokoro Dreams had unveiled in June 2005 the first commercial robot of theirs, Actroid DER (DER was described as meaning «Dramatic Entertainment Robot»). Actroid DER was created to be rented for exhibitions and other special events. Like all the other Actroids she is an air-powered robot, and this time, the robot is standing on a base with a movable platform under the left feet, allowing some leg movements. Actroid DER visited WIRED NextFest in the US in September-October 2006, later was used as a base for a student project of Carnegie Mellon University that lead for creation a modded version called Actroid Yume in May 2010, and to this day is available for rent as the younger model of the Actroid series.

    In October of 2006 Kokoro Dreams released the next model, Actroid DER2. The platform under the feet was removed, but realism of the appearance and the movement was boosted, with hands and feet especially noted as being the most realistic. It is arguably one of the best entertainment androids, used at numerous events in Japan and overseas (Campus Party in Spain in July 2008, Japan Day in New York in May 2009, Gitex in UAE in October 2015...). Over the years, the silicone face slightly degraded, Actroid DER2 «became older».

    Most of the performances were based on a pre-recorded monologue, but there were appearances where voice recognition was used, such as in New York on Japan Day, where Actroid DER2 answered questions about herself. Like the older sister, today Actroid DER 2 is still available for rent.

    Two year later on October 2008, Actroid DER 3 was released. This time another clever trick was used to animate the legs, first time used for Kokoro's one-time android in a shape of Holon from RD Sennou Chousashitsu — only one leg is attached to the base, with another one moving freely. Unfortunately, in many deployments this feature was underutilised, but with androids, it is quite common to see not all the hardware features used, a bit of an engineering sin, especially if you have in mind demoscene of the past, where people made sure that all the hardware futures of early computers would be used.

    Actroid DER3 is being rented to this day, and not only as a stand-alone unit, but as a whole exhibition about Actroids and android robotics in general. Kokoro Dreams continued making other robots, a simplified Actroid Sara in Novemeber 2009, a numerous amount of cleverly designed dinosaurs, a cartoony version of Charles Darwin, a mermaid robot for Thai pavilion on Expo 2012, another mermaid in the likeness of Actroid DER2, a robot in the look of rakugo performer Katsura Beichō, and even a curious lion dance robot with an android inside, which was exhibited alongside Actroid DER3 in P.R. China.

    One of the main issues with Actroids were the high prices of each unit prohibiting sales and long-term deployments. The air-powered nature also caused problems, as you could not just «plug in» the robot and call it a deployment. Multiple features such as leg movements, and even hand movements were not utilised fully, and so Kokoro Dreams had an idea to release a cheaper, electric-motor-driven Actroid, initially called Actroid SYR-KY-1.

    These robots were to be put as receptionists. The proof-of-concept for such a deployment first was shown using Actroid DER1 and Actroid DER2 on exhibitions. In February 2015, a hotel opened at the border of Huis Ten Bosch Dutch-themed theme park near Nagasaki. The hotel, bearing a name of Hen-Na-Hotel (there are different spellings, such as Henna Hotel), meaning «Strange Hotel» was staffed among other robots with receptionist robots by Kokoro Dreams, a dinosaur and a new Actroid. The hotel proved popular, and a few more hotels opened around Japan, including a couple in Tokyo, one of which i had visited. The hotel is a typical mid-high range business hotel, a bit overpriced maybe, but with all the geeky stuff working as it should be, including the Actroids which even impressed me by being able to read my Russian passport. Although i possibly shouldn't be too impressed, as passports are made to be machine-readable.

    Actroid SYR-KY-1 or just Actroid, was a successful concept, with robots appearing on multiple occasions in multiple deployments. Kyoei based their Receproid platform on the new Actroids. Actroid could be seen in spring of 2017 at the Narita Airport offering help with travel insurance on the departure floor. One Actroid since 2017 works at a recently opened Primetree Akaike shopping mall near Nagoya. These Actroids can be asked questions via a touch screen or by voice, although, i have seen voice recognition disabled, presumably, due to poor performance in noisy environments. Actroids both in Hen-Na-Hotel and in other places speak Japanese, English, Korean and Chinese, but sometimes only Japanese and English can be present. All the languages have existing TTX and voice recognition modules, so it all comes down to preparing the content.

    continued in part two ...

    Android Robotics up to 2019: The real story; in five parts; part 1; part 2; part 3; part 4; part 5
    Share post

    Comments 1

      «Not cake enough» звучит как приговор англоязычному хабру (который у меня теперь еще и по умолчанию выскакивает зачем-то).

      Only users with full accounts can post comments. Log in, please.