Willow Garage Blog

December 3, 2010


This week, Willow Garage hosted an NSF and JST supported workshop on Human-Robot Interaction, organized by Cynthia Breazeal (MIT), Fumihide Tanaka (University of Tsukuba), Brian Scassellati (Yale), and Leila Takayama (Willow Garage). We brought together researchers from the United States and Japan to establish and strengthen international collaboration across the Pacific, focusing upon remote presence.


Over two days, we shared our latest research, toured robotics research companies and Stanford labs, and discussed the future of remote presence. Our work covered a variety of robots, ranging from androids to mobile remote presence systems to humanoid robots to table-top expressive robots. It also cut across many industries, including education, healthcare and medical settings, socially assistive robotics, and geographically distributed work. By identifying common interests, fruitful application domains, and potential benefits and issues in the area of remote presence, we created some common ground for collaboration and a commitment to see these ideas through to development.


As a bonus, we also hosted a group of undergraduate students and post-doctoral researchers from Brown University's Robotics, Learning, and Autonomy Lab. They showed us their ROS-controlled quadrotor helicopter and flew it around the office. They also participated in a technical tour of Willow Garage, learned how to bring up a PR2 with ROS, and checked out our latest research and development demos.

All in all, it was a great couple of days for us. Thanks to everyone who came to Menlo Park to participate and made this such a collegial, engaging, inspiring workshop!
November 28, 2010


Sixty-five students spent the first week of November at the very first CoTeSys-ROS School on "Cognition-enabled Mobile Manipulation". These students focused on the challenges of personal robotics, like manipulating items in human environments. These students attended many lectures and also got hands-on experience with the PR2 robot and TUM-Rosie. They learned everything from basic ROS concepts, to navigation, perception, planning and grasping, knowledge processing, and reasoning. By the end of the week, they were able to program a robot to navigate to table, perceive objects on it, infer missing items, fetch the missing items, and bring them to the table -- an impressive feat.

The feedback on this first event was very positive. "Perhaps the most valuable was meeting researchers and other Ph.D students working in research areas closely related to my own, but of course the talks and tutorials themselves were almost as valuable; it would have taken much longer to learn all these things on my own," said one participant.

The event may have been limited to sixty-five students, but thanks to the organizers, you can watch videos of the lectures and download all the code materials. You can read more feedback on the event at the official report. Many thanks to the organizers in the IAS department at TUM and the many guest lecturers that made this first event a success.

The CoTeSys Fall School was jointly organized by IAS department at TUM (Technische Universität München) and Willow Garage

November 23, 2010

Bosch's AlanWhat is it about R2-D2 that makes him so popular?  Is it the cute-ish whistles, the grace under pressure, the slavish devotion to its owner, or just the fact that we surveyed a community of dedicated R2 builders?  Whatever it is, it's clearly working.  For R2-D2 has come out on top of our recent Favorite Fictional Robot Survey.  

Here are the final results:
1.    R2-D2, from Star Wars (81.5% 'Like It' votes)
2.    Bender, from Futurama (76.4%)
3.    WALL-E (73.9%)
4.    Marvin, in The Hitchhiker's Guide to the Galaxy (66.3%)
5.    HAL 9000, from 2001: A Space Odyssey (65.4%)
6.    Eve, from WALL-E (62.7%)
7.    T-800 or T-1000 from The Terminator (60.8%)
8.    Data, from Star Trek: Next Generation (60.1%)
9.    C-3PO, from Star Wars (58.9%)
10.    Johnny Five, from Short Circuit (53.3%)
11.    Optimus Prime, from Transformers (52%)
12.    Sonny, from I, Robot (49.8%)
13.    KITT, from Knight Rider (40.1%)
14.    Cylons, from Battlestar Galactica (39.9%)
15.    Rosie the Maid, in The Jetsons (27.8%)

As a reminder, we released this survey into the wild a few weeks ago to determine the favorite fictional robot from the Willow Garage community.  They survey focused on fictional robots (although we do appreciate the write-in votes for the PR2!) and was winnowed down to a Top 15 following an internal Willow Garage survey that included 40 original names.

While our survey methodology wouldn't pass academic muster, some interesting items emerged when we compared to the Willow Garage internal list. 

Survey results

The discrepancy between the top two spots is marginal, considering that the Willow Garage internal survey had Bender and R2-D2 neck and neck, albeit in a much smaller sample size.  What's most surprising is the big drop-off of Rosie.   Our guess is that our larger community skews a bit younger than Willow Garage employees, some of whom actually watched The Jetsons during the Kennedy Administration and not on-demand via Boomerang.

Poor Rosie was the least-liked robot by a wide margin; outdistancing KITT by a full 10 percentage points, followed by Optimus Prime. 

We also received nearly 200 write-in votes for our readers' favorite fictional robots.   Many of them were in our internal survey and didn't get enough support to make the public survey, but there were some great examples that we just missed.  All of these received multiple votes: the robot Andrew in Bicentennial Man; the robot servant GIR from the Invader Zim TV series; GLaDOS from the  video game Portal; the robots V.I.N.CENT ("Vital Information Necessary CENTralized"), B.O.B. (BiO-sanitation Battalion), and Maximilian from the movie The Black Hole were all mentioned; R. Daneel Olivaw, who appears is Isaac Asimov's Robot/Foundation Series.

Given the findings from the original internal survey, it's not surprising that Pee Wee Herman's Conky 2000 didn't receive a single write-in ;-)

The complete survey results are available here.

Image: Bosch RTC's "Alan" PR2

November 22, 2010

The ROS/Kinect integration continues to progress quickly thanks to the efforts of the OpenKinect and ROS community. The initial ROS+Kinect contributors -- Alex Trevor, Ivan Dryanovski, Stéphane Magnenat, and William Morris -- have combined their efforts into a new ROS kinect stack. At Willow Garage, our engineers and researchers have also been working on the stack to improve the driver and integrate it with ROS libraries and tools. The community is now hard at work on solving problems like calibration, which will be important for using the Kinect in robotics. Feel free to sign up for the ros-kinect mailing list to keep up-to-date on the latest efforts.

We thought we'd make a quick video to show some of what's going on at Willow Garage with the Kinect. We've added features like multi-camera support and control of the Kinect motors to the popular libfreenect library. We're also working on making some fun Kinect hacks of our own -- watch until the end of the video to see where we are with those. We look forward to seeing your videos as well.

November 20, 2010

Josh Tyler guest-blogged over at the WebMProject site about VP8 video running on the Texai remote presence platform.  VP8 is an open-source, BSD-licensed video codec released by Google.  It is also the best codec we looked at in terms of image quality, latency, and robustness to packet loss. 

Here's a short video that shows VP8 running on the Texai:

November 16, 2010

One of the hallmarks in human object perception is our ability to use a wide variety of sensory modalities. In contrast, most robots today rely almost exclusively on vision and/or 3D laser scan data for solving perception tasks. Many object properties (e.g., weight, material type, etc) cannot always be detected using vision alone - for example, visual feedback will not allow a robot to tell the difference between a full and empty bottles that otherwise look the same.

To address these types of problems, Jivko Sinapov from the Developmental Robotics Laboratory at Iowa State explored how proprioceptive sensory feedback, in the form of detected joint motor efforts, can be used by the PR2 for object perception. To use proprioception, the PR2 performed object exploratory behaviors such as lifting, unsupported holding, and sliding an object across the table. The robot learned a recognition model to detect whether an object (e.g., a bottle) is full or empty using features extracted from the joint efforts produced by the robot's behavior on the object. The robot was able to recognize whether a bottle is full or empty by lifting it from a tabletop, as well as by simply holding it in place.

These recognition models were tested with several different tasks. In one task, the PR2 had to solve a sorting task in which only empty bottles are cleared off a table. In another task, the PR2 had to estimate the weight of bottles. A third task -- sliding boxes across a table -- tested how quickly new recognition models could be learned. With minimal training experience (10-20 minutes), the robot was able to learn an accurate model for distinguishing between full and empty boxes.

For more information, please see Jivko's slides below (download pdf) or checkout the proprioception package on ROS.org.

November 15, 2010

Richard Hart came back to Willow Garage to do a news features for ABC 7 San Francisco. Last time he interviewed Dallas on a Texai Robot. This time around, he took PR2 and ROS for a spin, highlighting some of our hopes for the PR2 Beta Program.

November 8, 2010


It's hard to believe, but it has now been three years since we set out to create an open source software platform for the robotics industry. That effort has come to be known as ROS, which initially began as a collaboration between the STAIR project at Stanford and the Personal Robots Program at Willow Garage. Just a few short years later, we're excited to see how many individuals and institutions have joined in this collaboration. ROS (for Robot Operating System) is completely open source (BSD) and is now in use around the world in North America, Europe, Asia and Australia. There are robots running on ROS indoors and out, above and below the sea, and even flying overhead.

As we celebrate this occasion, we thought it would be a good time to share the "State of ROS" and talk about what's next.


There are now over 50 public ROS repositories contributing open-source libraries and tools for ROS, and more than 1600 public ROS packages. We are aware of over 50 robots using ROS, including mobile manipulators, quadrotors, cars, boats, space rovers, hobby platforms, and more.

People often ask how many users are there of ROS. Due to the free nature of ROS, we simply don't know. What we do know is that since the ROS C Turtle release this past August, there have been over 15,000 unique visitors to the ROS C Turtle installation instructions, with over 6,000 unique visitors in October.

ROS has grown very quickly this past year. Below are a few charts showing the growth in the number of public ROS repositories and ROS packages. As large as the ROS community is, you can see that things are just getting started.


Universities Using ROS

Academic contributors are the backbone of the ROS community, providing nearly three-fourths of the public ROS repositories. These contributions are helping to push the bleeding edge of ROS capabilities, and are also expanding ROS to new robot platforms. In the process, they are creating new communities within ROS to collaborate at the hardware, software, and research levels.

These individuals and departments have greatly expanded the range of hardware that can be used with ROS. Thanks to their efforts, you can now use ROS with iRobot Creates (Brown University), Aldebaran Naos (BrownUni Freiburg), AscTec quadrotors (CCNY), Barrett arms (MIT), Velodynes (UT Austin/ART), Pioneers (USCAMOR), Parrot AR.Drones (SIUE), and much, much more.

Several core ROS libraries are being built with the help of academic institutions: the ROS motion planning library is being built with contributions from the Kavraki Lab at Rice University, the CLMC Lab at USC, GRASP Lab at Penn, and Willow Garage; the Point Cloud Library is being built with contributions from researchers at TUMUni Freiburg, Willow Garage, and MIT; and the visual SLAM library is being built in collaboration with Uni Freiburg and Imperial College of London. There are also individual contributions: Jack O'Quin at UT Austin provided an official Firewire camera driver and René Ladan has done a FreeBSD port of ROS.

At the research level, the contributions are too broad to easily summarize. Whether you're doing research in 3D perception, manipulation, cognitive robotics, mapping, motion planning, controls, grasping, SLAM, HRI, or object recognition, there are ROS packages representing current research. We look forward to a world where "academic publication" refers to code as much as it does papers, and, thanks to the ROS community, we are starting to see that happen. We are also building tools to help researchers cite ROS code used in their publications.

Companies Using ROS

ROS has been adopted by robot hardware manufacturers, commercial research labs, and software companies. These institutions include MekaFraunhofer IPAShadow RobotYujin RobotThecorpora, Robotics Equipment Corporation (REC), SkybotixKuka (youBot), and Vanadium Labs. All are using ROS as a software platform for their customers to build on. Additionally, several of these companies are using ROS to build their own robot applications. ROS is also being used in research and development at Boschre2, Aptima, and Southwest Research Institute.

What's even more exciting than seeing companies using ROS is organizations contributing open source software to the community. There are ROS repositories for BoschShadow RobotAptimaFraunhofer IPA, Robotino (maintained by REC) and Vanadium LabsMeka and Skybotix are both providing open-source drivers for their hardware, and Thecorpora's Qbo is billed as an "open-source robot." More broadly, Yujin Robot has released their embedded tools for ROS and REC has ported ROS to Windows.

Companies developing software libraries for robotics have also been supportive of open source and ROS. Gostai completed the transition of the Urbi SDK to open source this year, and the 2.1 release added support for ROS. SRI released components from the Karto SLAM SDK as open source on code.ros.org and is supporting ROS integration.

Programs Using ROS

Various research programs are embracing ROS as a platform. ROS was created, in part, to support the PR2 Beta Program and encourage the exchange of ideas through software. This year, in addition to the official start of the PR2 Beta Program, there have been two DARPA programs announced that are using ROS: Maximum Mobility and Manipulation (M3) and ARM-S. The ARM-S program is providing a shared manipulation platform with ROS drivers that will enable participants to work with the ROS community.

And in Europe, there is the BRICS Project, which aims to identify and promote good development practice and reusable components for robotics. The BRICS participants are making use of many ROS packages, and working to integrate them with other robot software systems.

Building Bridges

One of the values of being open source is that it's much easier to collaborate than compete. With so much great open source software out there, it's wonderful that various robot software frameworks can build on each other's strengths rather than forcing users to choose between them. This year we've seen ROS integrated with OpenRTMUrbi, and PIXHAWK. There is also improved integration with Orocos.

ROS has also been integrated with other programming languages thanks to members of the community. Tim Niemueller has contributed a Lua client library for ROS, which also helps provide integration with the Fawkes Framework. Brown University has contributed a Javascript library for ROS that lets you control ROS-based robots directly from a Web browser.

Hobby and Low-Cost Platforms

ROS now runs on many lower-cost, hobby-friendly platforms. 2010 started off with Andrew Harris providing ROS libraries for the Arduino and was quickly followed by I Heart Robotics's WowWee Rovio drivers. You can now use Lego NXT robots with ROS as well as Taylor Veltrop's drivers for Roboard-equipped humanoids. Companies have also contributed: Vanadium Labs provided ROS drivers for their ArbotiX line of robocontrollers.

The ROS iRobot Create/Roomba community has also expanded greatly this year, with many institutions and individuals now providing drivers and libraries: Brown's RLABCU Boulder's Correll LabAptimaStanford, OTL, and ISR - University of Coimbra.

Making ROS More Open

We strive to make ROS as open as possible. From ROS's early days hosted on SourceForge, to a community-editable wiki, to open code reviews, we've done our best to perform ROS development out in full view of the public. However, we recognize that we can do more and are pursuing two major efforts to do so.

First, we created a new process for proposing changes to ROS. This process, called ROS Enhancement Proposals (REPS), empowers you to contribute to ROS development. It also provides better insight into current efforts.

Our second effort lays the groundwork for a potential ROS Foundation, an organization for the long-term development of ROS. We are inspired by the Mozilla FoundationApache Software Foundation, and GNOME Foundation, which act as stewards for public technologies. These foundations were not created overnight, nor were they created alone. We have already received invaluable advice from our friends at Mozilla on how to get started; now we need your help.

We invite you to get involved -- how can you play a role? We need developers for the core libraries, researchers to push the envelope, and companies to bring it together. As a community, these are all things we already have and are already doing. All we need to do is take the next step together.

This third year for ROS has shown us the size and strength of the ROS community. As our community continues to grow, we hope that we can better combine our strengths to meet the challenge of creating an open platform for robotics.

Concluding Thoughts

Just a few short years ago we set out to create an open software platform that lets roboticists focus on innovation, rather than reinventing the wheel; an open source robot operating system that is free for others to use, change and commercialize upon.

Three years later we are really excited by our community and what it has done. Whether you're talking about robots, libraries, companies, or research labs, the growth and breadth that we have seen has been stunning. We are grateful for your participation and have done our best to respond to your needs by making ROS better and more open. We're looking forward to the next three years (and many years after that), working together to build what's next.

October 29, 2010


In our line of work, there's a lot of debate about robot names.  For one, there's the whole notion of anthropomorphism and whether it's even appropriate to give a robot a name.  Setting aside that argument for the time being, if you agree that it is perfectly appropriate to give a piece of hardware, software and circuitry a name, then what exactly is the right name?

Our experience is that our PR2 users almost always give their robots a name, and that name is almost always based on a fictional robot.  So we asked the folks here at Willow Garage to tell us their favorite fictional robots.  That process gave us a very long list of names, so we have chosen the top 15 finishers.  Now we want the larger Willow Garage community to cast their votes.  Here are the candidates:

Bender, from Futurama
C-3PO, from Star Wars
Cylons, from Battlestar Galactica
Data, from Star Trek: Next Generation
Eve, from WALL-E
HAL 9000, from 2001: A Space Odyssey
Johnny Five, from Short Circuit
KITT, from Knight Rider
Marvin, in The Hitchhiker's Guide to the Galaxy
Optimus Prime, from Transformers
R2-D2, from Star Wars
Rosie the Maid, in The Jetsons
Sonny, from I, Robot
T-800 or T-1000 from The Terminator

Please cast your vote here and tell us your favorite fictional robot.  We'll let everyone know the results of our survey soon.

While don't want to influence the voting by letting you know how these top 15 fared against each other, we can tell you that Ash, from Alien, Deckard, from Blade Runner, and Voltron, from Voltron: Defender of the Universe just missed the cut.

At the other end of the spectrum was the Conky 2000 from Pee-Wee's Playhouse.  It received the least number of supporting votes, and a substantial number of 'hated it' votes.  (Something to do with the secret word, perhaps?)  Conky edged out Twiki, from Buck Rogers in the 25th Century and The Mechanical Hound, from Fahrenheit 451 for last place.

Let us know what you think by clicking here.  It will only take a few minutes of your time, but provide a lifetime of conversational fodder.

October 26, 2010

Jenay Beer from the Human Factors and Aging Laboratory at Georgia Tech interned with us this summer, exploring potential uses, benefits, concerns, and acceptance surrounding mobile remote presence systems for older adult users. We chose to focus on this audience as potential users because these technologies have the potential to aid in providing social support; potentially improving a one’s quality of life and helping to maintain independence.

This project was designed to answer a number of research questions:

  1. What are older adults’ views on what the system may be used for?
  2. What are older adults’ perceived benefits and concerns?
  3. Will older adults generally accept remote presence systems?

Jenay’s assessment focused on older adults interacting with the Texai. It was important that perspectives be based on actual experience as opposed to opinions elicited from people who merely imagine a remote presence “robot.” Twelve older adult volunteers worked with the Texai in two sessions. First, these individuals directly interacted with the system and conversed with a secondary researcher controlling the Texai. In another session, the volunteers were provided the opportunity to pilot the Texai themselves and communicate with the secondary researcher by controlling the system. Interviews with the volunteers followed each session.

There is a misconception that older adults will refuse to adopt new technology. In fact, the results from this study suggest the opposite. Overall, our volunteers were very positive toward the idea of using a Texai in their daily lives. In general, they identified many more benefits than concerns. These benefits were often related to maintaining one's independence. For example, the Texai video capabilities were considered more personal than a telephone. The Texai could potentially be used to go to events that would otherwise be difficult to attend (e.g., museums, live concerts, or sporting events). Additionally, the participants mentioned that they would like to use the Texai to contact friends and family, specifically their children and grandchildren.

Some overall concerns mentioned included the etiquette of refusing a call, and potential violation of privacy. The older adults recognized that a different set of social rules may apply to communicating through the Texai, in comparison to telephone communication.

Although older adults could see the general benefit of the Texai being used in healthcare applications (at a nursing home, for example), they also had concerns that the system could be misused or overused. In other words, they would like the Texai to be used as a healthcare supplement and not as a replacement for hands-on care.

Although the Texai are currently used in the workplace, this study provided a peek into this specific user group, older adults, who might benefit from using such mobile remote presence systems in the future.

It would not be surprising if these findings paralleled earlier research on another form of remote communication – the telephone. Could that technology be misused or overused? Would it replace face-to-face interaction or complement it? Would it be more readily adopted by specific demographics? Were older adults more averse to the introduction of the telephone than other age groups? Big questions deserve deeper answers, perhaps 424 pages worth. We’d encourage you to check out Claude S. Fischer’s America Calling: A Social History of the Telephone to 1940.