THE FUTURE IS ROBOTS•

Robotics Trends for 2012By Erico Guizzo and Travis Deyle

What’s in store for robotics in2012? Nearly a third of theyear is already behind us, butwe thought we’d spend some

time looking at the months ahead andmake some predictions about what’sgoing to be big in robotics; or, at least,what we think is going to be big. Lack-ing divine powers (or a time machine)to peek into the future, we had to relyon our experience as long-time ob-servers of the robotics landscape, cover-ing the field on two of the world’s toprobot blogs: IEEE Spectrum’s Automa-ton andHizook.com.

Tomake sure our forecasts aren’t toofar off, we asked a group of roboticistswith different backgrounds for their pre-dictions for 2012. This panel of expertsprovided invaluable insight and, afterwe tabulated everyone’s suggestions, wenarrowed it down to the final 12. It wasnot an easy task as there were so manygreat ideas. (Thanks, panelists!)

In making our selection, we triedto avoid the perennial trends—areassuch as environmental robotics (thetheme of this special issue), entertain-ment and toy robots, and others thatare always thriving with activity. Wefocused on emerging areas and we fol-lowed the money—looking at wherefunding is going. For example, the U.S.National Robotics Initiative (NRI) hasput a lot of resources into robots thatcan collaborate with people. TheDefense Advanced Research ProjectsAgency (DARPA), for its part, hasmultiple programs that involve manip-ulation and bionic devices. Europe’sFramework Programme is funding the

development of cognitive systems androbots that can assist people in every-day tasks. In Asia, the decade-longfunding of health-care robots for olderadults has intensified.

Again, we’re not trying to present acomprehensive survey of the state ofrobotic research in 2012. We’re hopingthat the trends described here are help-ful as a heat map that highlights promis-ing areas and technologies in roboticsand artificial intelligence. We know thatmany readers may disagree with ourchoices and that some might feel out-raged at our selections; others will justwant to learn more. In either case, wewant to hear from you. Write us or goto http://spectrum.ieee.org/robotics2012to see an expanded version of this articleand join the discussion.

As predictions go, only one thinglooks certain: robotics is going throughan amazing time, and things shouldonly get more exciting.

1) Co-Robots: Robots asCoworkers andCoinhabitantsTypically, robots fall some-where on a spectrumbetween direct teleopera-tion and full autonomy.Unfortunately, teleopera-tion can be cumbersomeand full autonomy is oftenillusive. Somewhere in themiddle lies a compellingtradeoff, where humansand co-robots can collabo-rate to perform practicaltasks (Figure 1). Co-robotsare at the heart of theUS$70-million NRI, spear-headed by theU.S. National

Science Foundation (NSF), and they rep-resent a definitive step toward robotsmigrating out of factories and academiclaboratories and into our everyday lives.According to the NRI, co-robots mustbe safe, relatively cheap, easy to use,available everywhere, and able to interactwith humans to leverage their relativestrengths in the planning and perform-ance of a task. “A lot of us are turningour attention in that direction,” said aresearcher from our panel of experts.(No wonder so many U.S. roboticistsspent the latter portion of 2011 draftingproposals!) The program’s scope isbroad, but the key aspect of co-robots isclear: they must interact with humans.So expect to see a big jump in activity inhuman–robot interaction in all itsmyriad forms throughout 2012.

2) 3-D Sensing: The KinectRevolution ContinuesLast year, a curious adornment startedappearing on the heads of manyrobots. It was Kinect (Figure 2), the

1070-9932/12/$31.00ª2012 IEEE MARCH 2012 • IEEE ROBOTICS & AUTOMATION MAGAZINE • 119

Digital Object Identifier 10.1109/MRA.2012.2184209

Date of publication: 26 March 2012

Figure 1. A study by Georgia Tech’s Human Factorsand Aging Laboratory used the PR2 robot to interactwith older adults in the Aware Home. (Photo courtesyof Keith Bujak.)

now-popular Microsoft three-dimen-sional (3-D) sensor. Being cheap andeasy to use, Kinect made 3-D mappingand motion sensing accessible, and therobotics community embraced it whole-heartedly. “People have been searchingfor a low-cost alternative to laser range-finders, and now we have one (forindoor use, at least),” one of the panelistscommented, adding that she expects tosee a surge in usage. Indeed, the Kinect2, whichmay appear sometime this year,will feature a higher resolution andframe rate that will allow the device, ifyou believe the rumors, to read lips. Newtypes of cameras also promise to expandthe possibilities of 3-D sensing. The so-called computational cameras—such asthe Lytro, based on technology devel-oped at Stanford—capture both inten-sity and angle of light and allow therefocusing of already-snapped picturesand the creation of 3-D images. Thenew wave of 3-D sensors may not onlygive robots better eyes but could alsoprovide an effective way of 3-D scanningeveryday objects, generating libraries thatrobots would access to finally understandthe thing we call the real world. “3-Dsensing is already hot,” said another oneof our researchers, “but with the Kinectand the next generation of similar cheapsensors, the sky is the limit.”

3) Cloud Robotics:The Forecast Calls for CloudsSeveral research groups are exploringthe idea of robots that rely on cloud-computing infrastructures to accessvast amounts of processing power anddata. This approach, which some arecalling cloud robotics, would allowrobots to offload compute-intensivetasks such as image processing andvoice recognition and even downloadnew skills instantly (see IEEE Robotics& Automation Magazine, vol. 18, no. 2,June 2011). A lot of activity should behappening in this area this year. As oneof the researchers said, “the cloud willexplode.” In particular, Google has asmall team creating robot-friendly cloudservices that, if they become popularamong roboticists, could cause a tectonicshift in the field (imagine every robotusing “GoogleMaps for Robots” for nav-igation). In Europe, a major project isRoboEarth, whose goal is to develop aWorld Wide Web for robots, a giantcloud-enabled database where robotscan share information about objects,environments, and tasks (Figure 3).Many other projects are taking shapeand we expect that, in 2012, cloudrobotics will sound less like a buzzwordandmore like a serious research domain.

4) Compliant Actuation:Robots with a Soft TouchWhen robots interact with humans,safety is a key concern. Conventionalposition-controlled arms, such as theones that dominate factories, justwon’t cut it. Making robots that havea soft touch is key to a future wherehumans and robots can share spaces

and collaborate closely. For this rea-son, we expect to see numerous im-provements to compliant actuationand tactile sensing technologies in2012. Examples of this include betterseries-elastic actuators (Figure 4) andtactile skin. In addition, we expectresearchers to think outside the box—

developing new, clever types of com-pliant systems far removed from elec-tromechanical motors. For example,in 2011 we saw a number of soft-bod-ied robots that, much like their meat-bag biological counterparts, are inher-ently squishy. These include iRobot’sHexapod JamBot, based on particle-jamming actuators, and OtherLab’sAnt-Roach Pneubot, an inflatable robotmade of fabric and pneumatic actua-tors. These and other projects point toa tantalizing future filled with compli-ant and soft-bodied robots that weexpect to start taking shape in 2012.

5) Smartphone-Based Robots:The New Robot BrainsAlmost every robot these days needs acombination of sensors, CPU, display,and network connectivity. Smart-phones and tablets offer a combina-tion of sensors, CPU, display, andnetwork connectivity. Do you seewhere this is going? At one end of thespectrum, iRobot has demonstrated aremote presence prototype robotcalled Ava, which uses a tablet to con-trol its mobile base. At the other endof the spectrum, two Seattle engineersquickly raised more than US$100,000on a fund-raising Web site, Kickstar-ter, with the promise of developing acute little smartphone-powered robot(Figure 5). These are just two

Figure 2. Microsoft’s Roborazzi robotfeatures a Kinect for navigation and acamera for snapping pictures of people.(Photo courtesy of Microsoft.)

Figure 3. A robot connected toRoboEarth serves a drink to a patientduring a trial of the system. (Photocourtesy of RoboEarth.org/TU Eindhoven.)

Figure 4. The Meka M1 compliant robotwith custom series-elastic actuators.(Photo courtesy of Meka Robotics.)

•

120 • IEEE ROBOTICS & AUTOMATION MAGAZINE • MARCH 2012

examples of a trend that we believehas an earth-shattering potential forrobotics. Mobile devices—based onApple’s iOS and Google’s Android—

are riding an extraordinary, unprece-dented wave of innovation. We think(and hope) that robotics can takeadvantage of this same wave. Theresult would be many more robotsmoving out of the laboratory and intothe marketplace. “More and moresmarts in cell phones, such as [Apple’svoice assistant] Siri, will impact robottoys and research this year,” one of ourpanel members commented. Indeed,thanks to smartphones, robots willonly get smarter.

6) Low-Cost Manipulation:A Robot Arm You Can AffordWith the ever-improving economicsof computing and developments suchas the popularization of 3-D sensors,the overall cost of a robot is droppingprecipitously. Except for one thing:actuators seem to be holding up theshow. This is most evident with grip-pers and high degree-of-freedom (DoF)arms for manipulation. We’re not talk-ing about hobby servo solutions; we’retalking about what many roboticistswant. They want a powerful system formobile manipulation: human-typeform factor, compliant actuation, re-spectable payload (at least 5 kg). Oh!Did we mention cheap? Like, all forless than US$5,000. In other words,could a new robot arm do for manipu-lation what the Kinect is doing for 3-Dsensing? The IEEE International Con-ference on Robotics and Automation

2011 paper by Stanford researchers ona low-cost compliant manipulatorsuggests that it is possible (Figure 6).Indeed, there is a fair amount of fund-ing in this direction—from the NRI ofthe NSF to DARPA’s three manipula-tion-centric programs [AutonomousRobotic Manipulation Hardware(ARM-H), ARM Software (ARM-S),and Maximum Mobility and Manipu-lation (M3)]. As one of the research-ers explained to us, all this fundingshould lead to “interesting new handdesigns and autonomous manipula-tion systems.” (Mobile manipulationwill be the theme for the June specialissue of IEEE Robotics & AutomationMagazine.)

7) Self-Driving Vehicles:Coming to a Street Near YouOkay, you probably won’t see autono-mous vehicles driving near you any-time soon; unless you live in SiliconValley, where Google has been exten-sively testing its famous self-drivingToyota Prius (Figure 7) and also afleet of autonomous golf carts. Butone thing’s for sure: autonomousvehicles have proliferated in the pastfew years, with projects in the UnitedStates, Germany, France, Italy, andChina. Last year, Nevada became thefirst U.S. state to permit autonomous

cars to be legally driven on publicroads (though some speculate thatEurope might prove more friendly tothis type of vehicle than the UnitedStates). Either way, autonomous driv-ing features are already showing upon regular mass-produced cars. Somemodels of the Prius currently have adriving-assist function that keeps thecar centered on its lane and anotherfunction that can park the car all byitself. Though carmakers will insistthat these are not autonomous drivingfeatures, it’s clear that cars are becom-ing more robotic. Furthermore, it’slikely that autonomous vehicles willdrive another trend as well. As one ofthe panelists explained, we should startto “map, and perhaps even instrument,our environment” to help autonomouscars and robots navigate. “This is ashift,” he said. “The emphasis used tobe solely on local algorithms and com-putation. Folks are starting to realizethat this is not the low-hanging fruit.”Driving, we’ll soon be able to say, is so20th century.

8) Factory Robot Helpers:The Future of ManufacturingLast year, an announcement fromelectronics manufacturer giant Fox-conn took the robotics community bysurprise. The Taiwanese company saidit was going to add 1 million robots toits assembly lines over the next threeyears. One million robots is a lot ofrobots—in fact, it’s double the currentindustrial robot population. Many un-certainties remain about Foxconn’splans, including whether they can pullit off. What’s clear, though, is thatthere’s a huge need for flexible, capable,and safe manufacturing robots—a

Figure 5. Romotive, a Seattle start-up,created a tracked mobile robot powered bya smartphone. (Photo courtesy of Romotive.)

Figure 6. Stanford researchers developeda low-cost compliant 7-DoF arm. (Photocourtesy of Morgan Quigley.)

Figure 7. Google demonstrates its self-driving car at the TED 2011 Conference.(Photo courtesy of Steve Jurvetson/Flickr.)

•

MARCH 2012 • IEEE ROBOTICS & AUTOMATION MAGAZINE • 121

new generation of industrial machinesdifferent from the big, expensive ma-nipulators in existence. Also, it lookslike this new generation is beginningto arrive. Examples include KawadaIndustries’ Nextage, ABB’s FriendlyRobot Industrial Dual Arm (FRIDA)(Figure 8), and Yaskawa’s MotomanSDA10D. But the factory robot thateveryone wants to see? It’s the mys-terious robot that Rodney Brooks isdeveloping at his supersecretive start-up Heartland Robotics. Is 2012 theyear he’ll unveil the machine? Wehope so.

9) Rapid Prototyping: A 3-DPrinter in Every HomeRapid prototyping is incredibly useful;being able to quickly fabricate a partcan save thousands of dollars, elimi-nate days of waiting, and allow you tofigure out whether your neat design isindeed brilliant—or a flop. There isone rapid prototyping device (a robotin its own right) that will likely makea significant impact in 2012: Maker-Bot Industries’ Thing-O-Matic 3-Dprinter (Figure 9). Three-dimensionalprinters are a mainstay of academicand industrial robotics laboratoriesthese days, but they’re expensive (somecost above US$20,000). The Thing-O-Matic (and its ilk) make 3-D printingavailable to the masses, retailing for aslittle as US$1,000. Low-cost 3-D print-ers could very well be the next bigtrend in home robots. MakerBot

forecasted that 10,000 units would besold in 2011, but we’re guessing thatthey have shattered their estimates sincethey were already ahead of schedule inMarch 2011 and raised US$10 millionin venture capital to expand their efforts.At the forefront of the do-it-yourself(DIY) maker movement, 3-D printerssuch as the MakerBot are fulfilling thepersonal fabricator scifi visions set forthin Neal Stephenson’s Diamond Age,allowing people to mock-up, share, andrefine digital products—usingWeb sitessuch as Shapeways and Thingiverse—for home fabrication. Our bet is that3-D printing is going to keep moving inone direction—up.

10) Unmanned Aerial Vehicles:Crowded SkiesSmall unmanned aerial vehicles(UAVs), and in particular, quadrotors(Figure 10), were huge in 2011. “Theyprovide an entry path to UAVs forpeople who couldn’t have worked onthem in the past,” said one of ourresearchers. “I would expect interest inthat to continue and grow.” Last yearwas just the beginning. In 2012, weexpect to see additional UAVs uptakeby both professional and citizenresearchers looking for inexpensiverobot platforms. Consider DIY Drones,

a popularWeb site for UAV enthusiasts;it boasts more than 20,000 memberswho design, build, and fly their ownautonomous UAVs. Buyers looking foralready-assembled models have lots ofoptions too. Walking through the mallduring the holiday season, we saw liter-ally a dozen stores selling UAVs, fromAir Swimmers remote-controlled (RC)blimps (below US$40), to small smart-phone-controlled helicopters (aboutUS$100), to Parrot’s ARDrone quadro-tor (US$300). Brisk holiday sales createdthousands of new UAV enthusiastswho, bolstered by ever-expanding opensoftware and hardware resources, willwork hand in hand with professionalresearchers to unveil some amazingflying machines in 2012. (The quadro-tor platform will be the theme for theSeptember special issue of IEEERobotics & AutomationMagazine.)

11) Telepresence Robots: YourAvatar in the Real WorldTelepresence robots—mobile machinesthat act as your stand-in at a remotelocation—first became prominent in2010, when Silicon Valley start-up Any-bots introduced one of the firstcommercial offerings, an alien-lookingrobot called QB (Figure 11). In 2011,more robots hit the market, includingthe Vgo from Vgo Communicationsand the Jazz from the French companyGostai. “It’s an important technologywith a strong business case and will saveboth time and money for business trav-elers,” said a researcher. Indeed, wethink that 2012 will be a milestone fortelepresence robots and robotics, ingeneral. By the end of the year, hun-dreds of QBs, Vgos, Jazzes, and others

Figure 8. The ABB FRIDA concept is a14-axis dual-arm robot designed forworking alongside human workers inmanufacturing environments. (Photocourtesy of ABB.)

Figure 9. One of the open-source 3-Dprinters created by MakerBot. (Photocourtesy of Bre Pettis/Flickr.)

Figure 10. A quadrotor helps assemble atower as part of an art project by ETHZurich roboticists and architects. (Photocourtesy of Markus Waibel.)

•

122 • IEEE ROBOTICS & AUTOMATION MAGAZINE • MARCH 2012

will be roaming around offices—a placewhere robots were nonexistent—all overthe world. It’s a first in robot history. In2012, a new entrant promises to makethis market even more competitive; it’slikely that Suitable Technologies, a Wil-low Garage spin-off, will introduce itsmuch-awaited remote presence systemas well. So expect to see more telepre-sence robots near you—if you don’tbecome one yourself.

12) Bionics: The Line BetweenHumans and Machines GetsBlurryCyborgs and other man–machinehybrids have long captured people’s

imaginations. We’re still far from thetechnology envisioned in science fic-tion shows such as “The Six MillionDollar Man” and “Robocop,” butresearchers have made significantprogress in the past two years. Areassuch as robotic prostheses and brain–

machine interfaces seem to be build-ing lots of momentum, and we expectto see some promising milestones in2012. In particular, exoskeletons areliterally strutting out of the labora-tory. This year, Ekso Bionics (for-merly Berkeley Bionics) will beginselling its robotic suit first to rehabclinics in the United States andEurope, hoping to have a model readyfor at-home physical therapy by themiddle of 2012 (Figure 12). At thesame time, a DARPA-sponsoredproject by Johns Hopkins Universityand the University of Pittsburgh hasbeen testing a brain implant thatallows patients to control an advancerobotic arm with their thoughts alone.Many other groups are also workingon technologies that promise to blurthe line between humans andmachines; it won’t happen overnight,but now the promise is not just sci-ence fiction anymore—it’s real.

AcknowledgmentsThe authors are thankful for thetimely and thoughtful feedback froma number of researchers: RaffaelloD’Andrea, Tiffany Chen, Matei

Ciocarlie, Steve Cousins, AaronEdsinger, Kaijen Hsiao, Charles C.Kemp, Masaaki Kumagai, MattMason, Hai Nguyen, Daniela Rus,Bruno Siciliano, Stefano Stramigioli,Gaurav Sukhatme, Russ Tedrake,Andrea Thomaz, and Holly Yanco.

Erico Guizzo, senior associate editor, IEEESpectrum. E-mail: e.guizzo@ieee.org.

Travis Deyle, founder of Hizook.comand postdoc fellow at Duke Univer-sity. E-mail: tdeyle@hizook.com.

Figure 12. A test pilot tries theexoskeleton created by Ekso Bionics.(Photo courtesy of Ekso Bionics.)

Figure 11. The QB telepresence robot byAnybots. (Photo courtesy of Randi Klett/IEEE Spectrum.)

•Turning Point (continued from page 128)

structures and mechanisms, such asbones, joints, muscles, and tendons.We’ve already built two robots, andnow we want to figure out how tocontrol them and give them human-like cognitive features.

EG: Is ECCE going to assemble mynext iPhone?RP: ECCE is a research platform. Ituses cables with a certain degree ofelasticity as tendons. There are 45motors embedded in its body that pullon the cables to make the body move.As for your future iPhone, I don’t

think it makes sense to put humanoidrobots in factories. The next-genera-tion factory robots could be just a pairof arms or hands, or other kinds ofmanipulators, which don’t even haveto look humanlike. However, maybe itcould be an ECCE hand.

EG: How do you control ECCE’smovements? How many degrees offreedom does it have?RP: Too many. That’s one of the big-gest challenges. For example, whenthe robot lifts one arm, the torso andthe other arm wiggle a bit. It’s very

biological and realistic in how itworks, but it doesn’t make controleasy. Lifting the arm requires theactuation of multiple muscles thatneed to be coordinated, actuated tovarying degrees. The way the bodyand muscles interact is difficult tomodel using classical control meth-ods. There are lots of nonlinearbehaviors, the tendons have static anddynamic friction, and the mechanicsare not precise. This is where learningbecomes critical. The robot needs tofigure things out by itself, at leastsome things. To be able to do that, we

•

MARCH 2012 • IEEE ROBOTICS & AUTOMATION MAGAZINE • 123