The following guest post comes from Gavan Gray, prolific commentator at JSW, who is currently a lecturer at Ritsumeikan University in Kyoto and a PhD Candidate with Leicester University’s Department of International Relations where he is researching developments in the Japanese defense industry.
The Skin and Bones of J.Robot
[This was written almost a year ago (19th June 2010) as an entry on my personal, non-public blog. More a journal for interesting news and trends than anything else. Seeing the recent post on Japan Security Watch wondering at the lack of Japanese robotics in the recovery efforts of the Tohoku disaster I thought my old notes might be of some interest to people curious about the state of robotics development in Japan.]
A recent article here has detailed the efforts of Geckosystems, a US Robotics developer, to pitch their latest products to the Japan Export Trade Organization (JETRO) and related groups. Among some of the points they made were their close ties to major Chinese companies and their interest in developing products for the commercial security and defense sectors. The two concerns that this should immediately raise are, firstly, the security of any data or proprietary technology being shared with Chinese companies and secondly, the legality of Japanese participation in joint development of products with possible security or military application under stringent application of the existing prohibitions on military exports. Both of these questions stem from the same failing though, namely that the Japanese government has not, as yet, sufficiently prioritized research into next-generation robotics as a national security concern.
While development of robotics in most other advanced nations inevitably has strong ties to their national defense branches, Japan’s legal and political restrictions have led to robotics being left largely to private universities and corporations with the sad result that there is no central body with the influence to support worthwhile projects and terminate the inane and impractical. The robotics industry has the potential to enter a dramatic growth spurt at any time. It is a technology on the cusp of widespread acceptance and once that moment occurs, the progression in technical capabilities and diversification of use is likely to be rapid. By then, however, it may be too late to institute the required changes to the current Japanese system.
As of the 2008 census there were some 8.6 million robots in the world, with a market value of $17 billion. While the market is expected to increase by 4% annually, by the end of this year the population is projected to double from this level as cheaper, mass-produced robots become more and more common.
The two leading applications of robotics thus far, are the industrial manufacturing and defense sectors. The US is the current leader in defense development and Japan is the world’s major industrial producer and user, with over industrial 370,000 robots on factory floors throughout the country. The utility of these robots are tied directly to economic health though, and with recent production in Japan falling almost 40% industrial robotics manufacturers have seen shipments fall by 59% in the past year. JETRO has previously shown its poor ability to forecast shifts in this market by predicting several years of steady growth while Toshima Koji, president of Yasakwa, Japan’s largest maker of industrial robots, puts things more accurately, saying simply “We’ve taken a hammering”.
Admittedly, JETRO did recognize one of the key problems when they said “For the next generation of robots we need to think more in terms of specific applications”. This is exactly the problem with the Japanese root industry at present. The vast majority of high profile creations are impractical exercises in simulation with no clear purpose. Even those designed for specific functions are aimed at niche markets rather than satisfying the needs that any ‘killer app’ must address. In order to break through a robot must be developed that is affordable, of clear benefit and use, and marketed as well as an iphone or pokemon.
But what areas should these be in? The aforementioned Industrial Manufacturing sector is clearly in stagnation although it will revive as soon as Japan’s economy does. Modern technology needs superhuman levels of precision and robots offer not only this but increased speed, strength, unlimited overtime and the ability to retrain a full workforce with a simple software upgrade.
Last year, a Heritage Foundation report broke down the primary areas of robotics development and, predictably, the Defense Industry was the other major growth field with demand in the US for military robots and unmanned drones increasing steadily in recent years. There are currently more than 6,000 robots on the ground in Iraq and Afghanistan with scores more circling the skies above them. Although some of them do possess offensive capabilities such as Qinetiq’s ‘MAARS’, the vast majority are for passive surveillance and bomb disposal purposes, areas within even the scope of Japan’s hamstrung constitutional limitations.
Many of the developments in this field are applicable to Policing and Border Security which has its own need for surveillance and bomb disposal. Mechanical contraband sniffers might also soon put an end to the sight of cute beagles winding their way between luggage at airports.
Another possible field is Disaster Response, with several foreign companies already producing machines, such as RoBug III, capable of exploring chemical and radioactive danger zones, and the swarm of Micro Air vehicles (MAV’s) created by the Swiss Institute of Technology, that can act as an emergency wireless network. Some progress has been made in Japan, such as at Tmsuk with their T-52 prototype and at Tohoku University where ‘snakebot’ can search for trapped survivors of earthquakes, but given the country’s vulnerability to natural disasters it is far less than might be expected.
Each of the above fields can make extensive use of UAV’s (Unmanned Aerial Vehicles). The most famous of these is probably the infamous Predator, an attack drone capable of firing high-explosive missiles from miles above the battlefield, but drones themselves have as much room for non-military application as aircraft. They are already employed extensively for Search & Rescue operations, for law enforcement and even for environmental monitoring. Japans Technical Research and Development Institute (TRDI) is conducting initial trials in both aerial and marine drones and given Japan’s high status as a leading developer of the sensors and integrated electronic systems required in these products it would be a serious error for Japan to fail to capitalize on their potential.
Medicine was the final field seen as ripe for expansion. By allowing doctors the access to robotic systems the potential exists for them to treat patients separated form them by significant distance. This allows specialists to have a far greater area of influence and for remote areas to gain access to much greater medical support. In the US there are already 200,000 robot-assisted surgeries each year using systems like daVinci, while the RP-7 diagnostic system allows for long-distance examination.
Looking at the five areas identified by the Heritage report though, we can honestly only place Japan as a leader in one, Industrial robotics, in the others, while there are some isolated projects and technical research and development, Japan cannot be said to be a leader. At the same time, other nations, including the USA, South Korea and upstarts like India and Sweden are making strong and sustained progress in the field. Should Japan remained unable or uninterested in seizing a share of these key markets, within the next decade it will be thoroughly eclipsed by countries with a better grasp of consumer needs.
So why does Japan wield such a strong reputation as a robotics leader. Apart from its indisputable industrial excellence its strength, and its fatal flaw, can be summed up in a single word – ‘anthropomorphism’. There is a fixation on designing robots that have ‘character’, ones that are cute, humanoid or helpful companions, so much so that frequently these factors are all they have to offer. The reality is that while such features may be a welcome addition at some later stage we first need cold, sterile machines that can actually help us solve problems. If such machines also happen to look like a cheerful Telly Tubby than so much the better but we need to make them do something practical first.
One of Japan’s first robot ‘hits’ was Sony’s Aibo electronic dog, which managed to sell 200,000 units before the plug was pulled on both it and Sony’s entire robotics division in 2006. Other colorful, and little more, contenders are Tmsuk’s samurai styled Kiyomori and their Hello-Kitty lookalike Rukuru, Kokoro’s lifelike Actroid and dainty iFairy, NEC’s PaPeRo, AIST’s Taizo, Toshiba’s ApriAttenda, Honda’s well-known Asimo and Mistubishi Heavy Industry’s (Japan’s top defense contractor) Wakamaru which failed to sell more than a few dozen units.
All the above have generally, very similar basic features. They are designed to interact with people, respond to questions, move and interact with simple objects. None of them do anything a human could not do much faster and more efficiently.
Tomio Sugiura, president of Sugiura Kikai Sekkei, a company that manufactures a vegetable slicing robot, talks wistfully of the day “every family would be having a humanoid robot”, but why the ‘humanoid’ clarifier. If the task requires a human form to do, it will be decades yet before machines can out perform humans and if not, why focus on the external shell?
There is certainly room for research into humanoid robots. Osaka University’s Child Robot with Biometric Body (CB2), the same university’s Gemenoid and the National Institute of Advanced Industrial Science and Technology’s HRP-4C are all examples of Japan leading the field in human emulation and, given Japan’s declining population, there may well be a place for them in the market of the future. Until that time though, there needs to be a significant increase in consumer demand for robotic products and this will not be filled with ideas that might be good given the twenty years evolution that even the most optimistic of robotics evangelists, like Ray Kurzweil, think are required before the machines can begin to match human intelligence. For the moment a plethora of failings, including, but not limited to, cost, safety, practicality and size constraints within Japanese homes, render them pipe dreams.
The success stories of Japanese robotics are the ugly siblings of these cute ‘helpers’. One armed giants like Yasakwa’s Motoman automotive engineering robots, or the box-like RFS-1, Fuji Heavy Industries office cleaner. They are perfect examples of function driving form and excel at what they do. Unsurprisingly this variety of robot ‘helper’ is the kind most other countries have focused on. The present global best-seller is iRobot’s Roomba vacuum cleaner. Essentially a smaller version of Fuji’s big box, it has already sold over 2.5 million units worldwide.
Other examples of not-so-cute but effective designs are: Kiva System’s inventory system, Floorbotic’s Floorbots and BelRobotic’s lawnmowers. While Japan (and South Korea) are focusing on appearance, other countries have invested more time on the artificial intelligence and pathfinding software need to get their target activity completed. One offshoot of this is that a number of distinct robotics development suites are emerging. These include Microsoft’s Robotics Studio (USA), National Instrument’s LabView (USA), Robuter’s robuBOX (France) and Gostai’s URBI (France). There are no equivalent Japanese software packages that are developing widespread acceptance.
It seems as though Japan’s disparate robotics developers work too much in isolation from one another with little overall guidance that could incorporate the best elements of each project into a superior gestalt. Frequently teams in other countries are linked by their sources of funding through either the government or defense sector. A recently formed Federation for Market Creation of Next-gen Robots (comprised of manufacturers Tmsuk, ZMP, vStone and BDL) believe Japan does not support robotics to same extent as South Korea (which is planning robot theme park and sets aside a $10 million annual fund for research). One more obvious difference is that participation by the Japanese Defense sector is extremely limited. Although military contractors such as MHI and FHI have developed robots of their own they have assiduously steered clear from defense applications, a serious failing given the important role such development has played in promoting the robotics industry elsewhere. Even iRobot’s best-selling Roomba would in all likelihood be discontinued where it not for the major sales the company also makes to the military.
It needs to be made clear that defense development does not automatically equate to offensive weaponry. The defense sector, especially in robotics, has as great a need for surveillance and support machines as it does new ways to destroy things. BigDog is an all-terrain pack-horse, Odis a bomb sweeper, Andros is an entire family of bomb disposal bots and Aeryon’s Scout is a tiny surveillance copter. These are only the bare tip of non-violent devices that could be developed in coordination with the defense sector and then applied to a variety of other commercial uses.
There are still major developments taking place in the field which show that we have not yet reached the true next-generation robotic systems, and thus, there is still time left for Japan to make core changes in its management of the industry. Researchers in Tokyo recently created a synthetic rubber skin that conducts electricity more than 450 times better than previous flexible coverings. The Ashikaga Institute has developed a wireless network for its robots that allows on the fly exchange of video feed without any cumbersome connecting cables. Several of Japan’s largest developers have formed a coalition to work towards allowing purely mental command of electronic devices and robotics. Elsewhere the Robots and Mechanisms lab at Virginia Tech has developed robot cars, the University of Massachusetts is making advances in telepresence and Sweden’s iRobis is pushing the boundaries of evolutionary AI. Any one of these projects could mark a turning point in development of that ‘killer app’ that seizes the public consciousness and truly marks widespread acceptance of the utility of robots.
In order to avoid missing the window Japan has to seriously reprioritize its research focus. It should start by clamping down on further efforts to make ‘cute’ home helpers that have zero sales potential, and rejecting aimless ideas such as the Japanese Space Headquarters plan to have a two-legged robot walk on the moon. METI’s Robot Industry Policy Committee is continuing to advocate robots as primarily for elderly care, seemingly blind to the significant limitations of their scope.
Which leads to the second vital change. There needs to establish a clear central agency with oversight of technology that is vital to national security. Is the above RIPC the central group or perhaps the New Energy and Industrial Development Organization? Or one of several other groups with fingers in the robotics pie such as the aforementioned next-gen Federation? The government needs to select one and give it more teeth to coordinate research, development and marketing for the industry. One of its first actions should be to redraw the targets for Japanese research and stipulate the need for practical systems in order to gain support and funding. A second would be to fast-track the development of safety standards for robots so that Trade Ministry laws preventing robots and humans from working together can be redrawn.
Next, there needs to be greater integration between the Defense sector and development teams. While the government’s TRDI is already working on robotic and drone systems, it needs stronger links to the manufacturers such as Mitusbishi and NEC who can make these things a reality. Technical development without clear market research, and reconsideration of the impact of export prohibitions, might simply result in cutting edge technology (like Mitsubishi’s Wakamaru) without a single buyer.
Finally, the government needs to solicit greater input from both the Japanese public, private institutions and international development teams. Whether simply ideas or working prototypes. Competition such as Australia’s MAGIC 2010 reconnaissance challenge and the US NIST/DHS annual Response Robot Evaluation act as proving grounds for some of the best systems in the world and help cut the wheat from the chafe. METI currently hold annual robot awards for outstanding developments in certain fields but should initiate similar programs that drive innovation rather than simply respond to it.
It may seem like a lot to ask of the Robotics Industry but given that what is at stake is the nation’s self-preservation as a leading developer, it is undeniably necessity. All that it takes is a strong and resolute stance by the government with initiatives to support more practical application of research. It is either this or a continuation of mawkish ‘humanoids’ that reveal the true depths of the uncanny valley and home ‘help’ that is likely to either cripple its elderly owner when it falls on her or remain unused in a storage closet. Such projects no doubt have their place in helping introduce robotics to the public but they are merely the surface skin on an industry that needs a far stronger skeleton at its core if it hopes to prosper.
A former contributor to World Intelligence (Japan Military Review), James Simpson joined Japan Security Watch in 2011, migrating with his blog Defending Japan. He has a Masters in Security Studies from Aberystwyth University and is currently living in Kawasaki, Japan.
His primary interests include the so-called 'normalization' of Japanese security (i.e. militarization), and the political impact of the abduction issue with North Korea.
James Simpson has 254 post(s) on Japan Security Watch