By Frank Tobe, editor/publisher, The Robot Report
Yesterday was the launch of Rethink Robotics' Baxter shop assistant robot.
Many reporters, including myself, saw Baxter at Rethink's headquarters in Boston in the past two months but were embargoed from writing about it until today, the launch date.
Looking at the stories - as they appear all over the web and in news print, three stand out, in my opinion, at the head of the class:
- the Rethink Robotics press release
- the NY Times article and video by John Markoff
- and the in-depth story in Spectrum by Ackerman and Guizzo
As an aside, Markoff's article makes me ever more aware of the difference between a blogger/writer and a paid reporter from a reputable news source. John's article made me wish I could write like him. I saw what he saw, got the same demo, played with the machine similarly, heard the same details. But I came away disappointed.
- John saw, as I did, that both arms don't work together. He wrote that they will in the next version. I couldn't imagine them bringing the product out without the arms working together (although they are aware of each other's location and can avoid hitting itself).
- John saw, as I did, that it was big but he just described it as a 9' span that, with the stand, brings Baxter eye-level with a standing worker. I saw it as big and clunky.
- John heard, as I did, that the price was $22,000 yet Brooks had been touting an affordable cost of less than $15,000 for a long time and cited polls that said that the price point for new shop tools was $15,000 or less. John just wrote the facts and let it go at that.
- We both saw that the grasping mechanism was a two-fingered plastic device and heard that, in the future, third party mechanisms could be quickly retooled to enable Baxter to do different tasks. I thought this was an awkward solution and found that Baxter could only lift 5 pounds and didn't have enough torque to push/screw in a screw which, in my opinion, would be a necessary task for any shop assistant. John just showed what it could do with it's two fingers.
- John described the need of small shops to augment their workers with robots which can do the dull, repetitive parts of their daily work much easier, thereby freeing up the worker to do higher-level tasks. But he didn't indicate that this version of Baxter is far away from offering that capability. Nor did he talk about other promises and expectations missing from this launch, namely an app store where Baxter's training can be transferred to other places where other Baxters are installed.
- We both saw that Baxter's speed was slow and its precision not too precise but John just reported what he saw; I wondered how long it would take to speed up the processing and enhance the mechanisms to be more precise.
Markoff only writes a robotics story every month or so; Ackerman and Guizzo write every week or so. But every day there are hundreds of stories that need sifting to see whether they fall into News or Views and if News, whether the article is applicable for our readers as they do as we do: track the business of robotics.
I wrote this meandering message because of a quote I think appropriate about Baxter: George Carlin, the late comedian said: "Inside every cynical person, there is a disappointed idealist."
***
Other stories about Baxter and Rethink Robotics from reliable sources:
- Bloomberg Businessweek: Smarter Robots, With No Pesky Uprisings
- Technology Review: Meet the Other Robots Set to Invade Manufacturing
- The Wall Street Journal: Baxter Robot Heads to Work
- PC Magazine: Meet Baxter the Robot, Your Potential New Co-Worker
- Forbes: Manufacturing Gets A Makeover With Baxter
- Popular Mechanics: It's Time To Welcome Our Robot Underlings, Starting With Baxter
- The Engineer: Rethink Robotics Unveils Humanoid Factory Robot
George Carlin
ReplyDeleteFixed. Thanks.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteGreat write-up. Humble and informative. I especially enjoyed the meta-level commentary about journalism vs. blogging. Thanks, Frank.
ReplyDeleteSorry, but I disagree with the previous commenter. This post is quite clogged with your commentary on how you can't write as well as Markoff and doesn't tell us much. As someone who studies this industry a lot, you probably have some insightful things to say about the robot. I'd love for you to write about them more clearly where they aren't confused by thoughts on writing styles.
ReplyDeleteFor instance: Why are two arms working together important? What kinds of tasks does that preclude? Don't most industrial arms work alone? Why is 22k vs 15k such a big difference? Isn't that still an order of magnitude lower than most robots and still going to open up robotics to those who previously couldn't afford it?
22k vs 15k is a huge difference because entry-level industrial arms start at around 20k from the more premium manufacturers. Entry-level arms from a budget manufacturer are around 15k and still offer a much greater functionality than what Rethink has displayed in Baxter.
ReplyDeletehttp://www.youtube.com/watch?v=jRbr8ielrOI&feature=player_detailpage
ReplyDeleteIn 2009 showen Man-Robot-Interaction and Cooperation with KUKA Robotics LBR (Light Weight Robot); working today at Mercedes Benz.
... and do not forget to multiply the cost of Baxter by ten to arrive at the cost of the KUKA/DLR _LBR.
DeleteIn our laboratory automation community, we view Baxter with interest as the feature set evolves. Baxter appears to be a broadly general purpose robot aiming to mimic human operations in a production environment. Most of the robot applications in laboratory automation are purpose built, typically handling a specific set of labware. The precision requirements have increased over the years, as have speed requirements.
ReplyDeleteIn 1981 when Zymark introduced the first laboratory robot it was a general purpose robot within the specifications of that generation's laboratory operations. We were handling large containers from flasks to 50ml centrifuge tubes to test tubes. Throughput needs were medium and batch sizes were relatively small. By the end of the 1990s High Throughput Screening became a dominant force in laboratory automation. HTS requirement for speed, handling precision, large sample sets and a low error rate drove the industry to more industrialized purpose built robotics.
Your article and John Markoff's NY Times article pointed out some of the current challenges with Baxter. The arms must be coordinated to work together - this is a critical issue for laboratory operations like capping/decapping and liquid handling. Speed and size should be addressed. As mentioned, the labware handling must be done quickly. Real estate in most labs is at a premium. It does seem that Baxter's design is amenable to scale-down. End effectors need to be robust and precise, and there should be the ability to exchange graspers for different tasks.
That being said, the price point is extremely attractive compared to current offerings in laboratory automation. I see a fit in many labs who would like to utilize automation but find present offerings too expensive, particularly those labs with medium throughput requirements. As well, the simple instruction set means that an organization could effect automation with reduced support from outside contractors. I believe we will see Baxter and kin in laboratory automation as the design improves.
Frank, I think this article is great. We need more realistic and truthful evaluations of so called "robotic products" and less PR spin. All too many times I see articles from blogs and news agencies that proclaim this "robot" or that "robot" is the greatest thing since slice bread. Baxter is a neat idea. Its unfortunate that a majority of its IP comes straight out of the CSAIL lab and other ppls work. Its design is marginally interesting compared to other similar systems on the market today. What I think is the biggest unmentioned news about all of this is how Rod (and Rethink) are controlling the IP on SEA actuator technologies and not allowing any other companies to sell products into market (outside of R&D at labs and schools). This is and will continue to be the biggest part of all of this.
ReplyDeleteThe SEA patent was 1995. Only 2 years until it expires. It will be interesting to see how Redwood Robotics deals with that fact. They're a joint venture with Meka and all of Meka's products are SEA based.
ReplyDeleteYou are correct about that. The big "BUT" is how they can re-apply for protection on it, with respect to, its usage in a system (product). The original patent protects the core technology (control and mechanical implementation). An example of this is the packbot wheel. The mechanism itself was (and is) used in many areas of technology (look at every rover every sent into space) but its usage on a mobile robot is protected because of the original Urban/packbot patent on it.
DeleteAnother thing to keep in mind is how Rethink is creating their gear reductions for the arms. Its not harmonic drives, nor is it any typical right-angled gearing. It is either a traditional planetary (highly unlikely) or a split ring planetary design (very likely) (also know as the "cheap mans harmonic" or fixed-differential drive) Its been used in winchs for many years. They could, if they wanted to, protect that as well, even though its a known design in industry.
There is another solution for SEA based product, patent #8209052 : High performance differential actuator for robotic interaction tasks.
ReplyDeleteHere is the link : http://www.google.com/patents/US8209052. This is from the IntRoLab robotics lab where I work at Université de Sherbrooke, Québec, Canada. More information on the lab here : https://introlab.3it.usherbrooke.ca/mediawiki-introlab/index.php/ADE.
Dominic,
ReplyDeleteNeat. Because the spring is tied to a mechanical element in parallel with the load this differs enough to be patent-able? (i.e. not "serial"?) This is essentially what I had in mind with my previous point of "fixed differential drive". Where the "fixed ring gear" is torsion-ally locked to the frame by a spring. Whether you apply the design through a harmonic or other type of gearing.
Please be aware that there are currently a manifold of approaches extending and going beyond the SEA concepts known under VIA (variable impedance actutors). For example, please have a look at http://www.viactors.org/VSA%20data%20sheets.htm .
ReplyDeleteHi to all,I am new person to the blogging stuff.The content in the blog is very informative and effective one.Keep update more things about the robotics.
ReplyDeletePongRus