This is a big deal. After $62M in investment and a commercial development effort extending back to 2008 (earlier if academic research work is also considered), on September 18th Rethink Robotics announced its first product offering, a new class of robotics technology designed primarily for small-to-medium manufacturers.

The robotics and automation community had not witnessed this level of breathless anticipation for a new product since the release of the Segway Personal Transporter in 2001. The excitement was not limited to the robotics community only. Investors, business development groups, government officials and others were keenly interested in what Rethink released. For me personally, the Rethink announcement was the equivalent of Apple’s initial iPhone and iPad releases, with LeBron James’ “The Decision” thrown in for good measure.

Over the years the high level of interest in Rethink Robotics has resulted in a great deal of coverage in robotics and business media outlets. I was not immune to the enthusiasm, and in a June 2012 Robotics Business Review article I speculated as to Rethink’s product plans. In the large I was correct, but big picture analysis is easy.

The details of the Rethink announcement reveal a unique technology and a compelling business model, which is expected to revolutionize manufacturing, and, I believe, have substantial social and political ramifications. Therefore, for this article a technical specs review and brief product description alone simply will not suffice (you can find that aplenty on the Web, along with reviews of the company’s experienced management team). For roboticists, the investment community and manufacturers, it is important to understand Rethink’s product and business plans, but also the wider industrial robotics sector and the social and political consequences of this unique product launch. That is what is attempted here.

Baxter in Brief

Rethink has announced the availability beginning October 1st of Baxter, a low cost, dual armed humanoid robot primarily targeted to small-to-medium manufacturers. Each of Baxter’s 7-degree-of-freedom (DoF) arms provide a rated payload of 5 lbs. The human scale robot will come loaded with a variety of sensors out of the box, including vision, touch, force and position. Baxter, which can be set up and reconfigured quickly and easily, is designed to work collaboratively and safely with humans. In addition, Baxter can be programmed by non-technologists.

The thumbnail description above is accurate as far as it goes, but it does not do justice to the scope of Rethink’s vision or what the release means in the larger sense. That requires, as a first step, an understanding of the genesis of the Baxter system, including some insights into the industrial robotics market sector.

SMB Manufacturers a Missed Opportunity

Traditional industrial robots have been successful in manufacturing because they could quickly, repeatedly and tirelessly perform a task such as welding and painting, often under extreme conditions. The benefits of such automation are many including:

• Increased product quality
• Improved capacity utilization
• Improved labor efficiency
• Improved worker safety (frees workers from performing tasks hazardous to their health)

For the most part, the industrial robotics sector, which reached maturity over the second half of the last century, consists of immobile, single task robots that have little interaction with humans or the world around them as they perform their simple, repetitive tasks while encaged within a safety barrier of some type. Not only are the robots themselves dangerous, they usually perform hazardous or unhealthy tasks.

Automotive and electronics manufacturing account for the overwhelming majority of industrial robotics installations. Even when robots made inroads into other markets, say, the pharmaceutical industry, they were purchased by large enterprises that could afford sizable amounts of capital investment and had large production volume requirements.

Small-to-medium businesses (SMBs) lack sufficient funds for expensive robotic installations. They require low volume, flexible automation solutions that can easily be changed to meet shifting demand. In addition, many SMBs do not have the personnel available with a deep understanding of how robots operate or how to program and maintain them, nor can they afford to outsource or use robotics consultants for an extended period of time. As a result, small-to-medium manufacturers have largely missed out on the benefits of robotic industrial automation described above, despite the fact that the market for robotics installations into small-to-medium-businesses exceeds that for large enterprises.

Designing for SMBs

Industrial robotics manufacturers are aware of their own sector dependence, as well as the size of the market represented by small-to-medium businesses. Academics, often working with governmental agencies, are also interested in the market, particularly as a means to increase international competitiveness and promote job growth (Europe’s SMErobot Initiative provides one example). Taken together, these groups have come to a rough consensus as to the features and functionality of robotic automation technology optimized for SMBs (again, particularly manufacturing). These include systems that:

• Satisfy the requirements and support the culture of SMB manufacturers
• Support manufacturing that makes no assumptions as to volume levels or even types of products
• Support multiple types of automation tasks
• Are lightweight, portable and offer a small footprint
• Come in at a low purchase price and dramatically reduced life-cycle costs
• Offer high reliability and easy maintenance
• Provide quantifiable cost justification and short payback
• Allow for rapid deployment and integration
• Work safely and effectively in workspaces occupied by humans
• Can be programmed easily and quickly
• Can be operated by a workforce with a wide range of education and qualification levels
• Have multiple options for purchase financing and robot use (rentals, for example)

Mixed Results

Efforts have been made to develop robotics systems suitable for SMBs, but the results have been mixed. Many academic efforts have resulted in prototypes, but commercial systems are lacking. Producers of industrial robots have reworked current models, reducing system size and cost. But the repurposed, smaller versions of existing systems often do not meet the requirements of SMBs or mesh with their cultures. Systems remain a challenge to program, requiring engineering expertise and the use of PCs and teach pendants, hand-held robot control terminals that provide a means to move the robot, teach locations and run programs.

Even with reduced purchase pricing, the total cost of ownership for industrial robots is simply beyond the means of many SMBs. The capitol costs for the robots themselves account for only 25%-30% of the total system costs. The remaining costs are associated with robot programming, setup, and dedicated, shielded work cells. The relative impact of these costs is exacerbated for companies requiring constantly changing, small batch production runs, exactly the type of manufacturing required by the majority of SMBs.

What small-to-medium manufacturers require are industrial robots that are specifically tailored to the SMB market. The systems should be low cost, simple to set up and reconfigure, programmable by non-technologists and able to work safely without a dedicated space or specialized equipment. Enter Baxter.

A New Class of Automation Technology

It would be a disservice to describe Baxter simply as an industrial robot, even as one engineered particularly for SMBs. Industrial partner robot might be a more apt term. This is more than a semantic distinction. Rethink believes, and I would agree, that Baxter is a new, even radically new, class of automation technology, and one that does not compete directly with traditional industrial robots, but instead acts as a complement. Baxter will be used in a wholly new ways and will greatly expand the types of manufacturing tasks that can be automated.

Rethink Robotics representatives describe Baxter’s target market as small-to-medium manufacturers, although larger manufacturers are free to employ the system. This alone does not qualify Baxter as a new category. However, a number of other factors, taken together, does:

• Out of the Box Usability: Baxter is made to be used right out of the box. That is, there are no additional charges or requirements for camera systems, safety equipment and the like. Also, the system can be set up and put into production very quickly.
• Programming Ease: Baxter can be used without programming in the normal sense. The system is preprogrammed to perform certain tasks, and then the robot can be trained further to perform work much in the way a human worker would be trained. For example, Baxter’s arms can be physically positioned and objects recognized and grasped to simulate a task. Once a sequence of such actions are entered and confirmed by the robot (with a nod of the robot’s head), Baxter is free to work.
  For more complicated tasks, Rethink also supplies an SDK providing access to all system functions. This manner of floor-to-ceiling programmatic access, where functionality is highly abstracted, but can also be accessed at a very low level, allows Baxter to be programmed easily, but yet places no limits on the types of programs that can be developed. This in itself provides a type of investment protection.
• Low Cost: As fitting for the SBM market, Baxter is relatively low cost, coming in at a list price of $22K. Rethink does offer a parallel gripper for $1,200, as well as a vacuum cup system. An optional pedestal including casters is also available. A complete, fully functional, dual arm Baxter system with pedestal and gripper can be had for approximately $27K. This price compares favorably with even traditional single arm systems (cameras, sensors, grippers etc. must be purchased separately).
  Baxter comes with a 1 year warranty, and a 3 year extension is available. Software upgrades are offered for one year at no cost. A software subscription service is under consideration, but no decision has been made at launch time. Rethink representatives have noted that they intend to continue drive down costs and are also willing to work with 3rd parties to provide additional Baxter compliant accessories.
• Safe Operation: Rethink’s system has been engineered to work safely in close proximity to human workers, and does not require safety cages or other specialized equipment. Baxter boasts of compliant joints and fully backdrivable motors in the arms. Onboard sensors monitor for human proximately and collisions, with the robot adjusting its actions as necessary (ex. slowing down, stopping etc.).
• Adaptive, Flexible Automation: Rethink’s Baxter is designed as a flexible automation platform that can be repurposed quickly and easily to suit needs. This aligns with the requirements of SMB manufacturers, as well as the latest manufacturing methodologies such as Adaptive or Agile Manufacturing, where production runs change constantly and production volumes vary widely.
  This approach contrasts sharply with many classes of industrial robots that are purchased and used task specifically (painting robots, for example). In other cases, generic industrial arms must be augmented with sensing systems, complex programming and specialized workspaces limiting their applicability to a narrow set fixed manufacturing tasks.
• Extensible Platform: Rethink describes Baxter as an extensible platform. In this case extensibility takes on two aspects. First, and more traditionally, the robot can be purposed to a wide variety of tasks (see Horizontal and Vertical Markets, below). As noted previously, the system can also be used for certain classes of manufacturing processes immediately with minimal robot training. The available SDK allows for programming more complex tasks. These are examples of technical extensibility.
  “Extensible” can also describe Rethink’s business model, which in many ways is atypical of a producer of industrial robots, and more in keeping with makers of personal computers (according to Rethink representatives), and even developers of software for mobile devices (my own opinion).

Atypical Business Model

Rethink’s funding chain followed a formula much in keeping with other high tech ventures’ Series A funding is followed by Series B, and then a final go-to-market round. With the Baxter product release, however, Rethink colors outside the lines. First, while the Baxter hardware begins sales on October 1st, the product will ship with beta software. The release date for the final Baxter code is January 1st, 2013 (happy holidays Rethink software engineers!).

Rethink’s software approach approximates the Lean Startup Methodology that has been embraced by developers of Web centric software and apps for mobile devices. Lean Startup advocates releasing a “minimum viable product” to early adopters that can be quickly validated. The product is adjusted as necessary based on customer feedback, rapidly released again, and this process is iterated so as to achieve optimal “product market fit”. This approach offers the added advantage of releasing product sooner and generating revenue immediately.

Rethink representatives describes their software model in terms of the PC and workstation market. Accordingly, software will be released on a regular basis, with each new release providing for new tasks and offering better performance. The difference between the Lean and PC models is the maturity of the initial software runs and regularity of releases.

Platform Play

Regardless of how the approach is framed, it is clear the Baxter is really a platform play. Hardware advances will be incorporated into the Baxter platform, of course, and more ancillary products made available, including through 3rd party hardware partners (grippers and other types of end effectors, for example). However, competitive differentiation based on low cost sensing technology and commodity hardware (motors etc.) is difficult to achieve. Baxter the software platform, or possibly the software stack (or even ecosystem if outside partners are involved), is another matter. Here competitive advantage and differentiation can be had more readily, and IP protection more easily enforced. In conversations with Rethink representatives, it was made abundantly clear that the company plans to aggressively protect its intellectual property.

Horizontal and Vertical Markets

As noted above, Rethink is initially focusing on small-to-medium manufacturers as their target market. No specific vertical markets have been targeted, as Baxter has wide applicability. According to Rethink representatives, the initial system release is well suited for light material handling processes such as loading, unloading, placing, or manipulating material for manufacturing operations such as part transfer and machine tending. The company expects that additional classes of operations can be performed in the future as new software and grippers become available (from both Rethink and 3rd party developers). Example Baxter manufacturing operations include:

• Pick and Place : Baxter is well suited for the repetitive part transfer operation of loading and unloading items on a moving assembly line (ex. picking items off a conveyer and placing them in a pile, array or container and vice versa).
• Machine Operation/Tending: Baxter can be trained to load and unload raw materials into machinery for processing, or start and stop a process.
• Part Testing and Sorting: Baxter can be programmed to test or sort random objects on an assembly line as part of quality control or other need.
• Packing: The Rethink system can pick and place components into containers for shipping or storage.
• Light Assembly: Using both arms in coordination, and likely with a human counterpart, Baxter can be programmed to perform some types of light assembly work.
• Finishing Operations: Baxter can be used for light finishing work such as removing excess plastic or metal from parts.

Market Sizing, New Markets

According to Rethink representatives, there are approximately 12M manual labor manufacturing jobs in North America. A back of the envelope elimination analysis, say by removing jobs in specialized vertical markets such as food production, work involving heavy loads, or assembly tasks requiring high levels of skill and fine manipulation, reduces the 12M figure substantially. Still, there are approximately 800,000 manual labor jobs in general manufacturing where light payloads are held with one hand and are involved in simple and repetitive non-skilled tasks.

Rethink pegs the size of the North American market for automating these classes of processes with Baxter-like automation technology at $16B. The worldwide market, obviously, is much larger. For comparison (albeit imperfect), the International Federation of Robotics (IFR) calculates the size of industrial robotics sector worldwide in 2011 to be approximately $8.5B (not including systems engineering, software etc.).

While market sizing, especially where no existing market exists, is an inexact science, the perils of prognostication can be reduced by market understanding and using correct assumptions. Rethink appears to have done this. But at some level, the actual figures do not mean all that much.

What does matter is the knowledge that much of the manufacturing sector has not been automated at all. Even more importantly, low cost, very powerful, commodity sensing technology is now available and at one time it did not exist (at least at a reasonable price point). These sensors, combined with increasing powerful (and low cost) processors, can be harnessed by smart software to create new classes of affordable robotics systems, including manufacturing systems. New products, as well as new markets will be created by those who understand the robotics technology power/cost dynamic and apply it in novel ways. This is what Rethink has done.

Much in a Name

Human faces as a means of communication have been proven in the field for over 200,000 years, millions of years if prehumans are considered. The most striking difference between Baxter and the majority of industrial robots on the market is its humanoid form factor. The robot has a distinct torso sporting two arms, and a separate head with a LCD graphical face including two eyes. Baxter is proportionally similar to a human, and is roughly the height of a human male, standing 6′ 1″ (1.8M) when the optional pedestal is employed (the torso alone is 3′ 1″ – 94 cm).

Rethink named its robot “Baxter”, a telling departure from the usual naming conventions for industrial robotics products. Consider KUKA’s KR 1000 “Titan”, FANUC Robotics’ M-430iA, Adept’s Quattro s650 and Motoman?s SDA10 (also duel armed). These names are by no means unusual for industrial robots that as a rule sound impersonal, industrial (naturally) and slightly aggressive.

As a given name, Baxter, meaning “baker” in the original Anglo-Saxon, is nonthreatening, and connotes humor and even playfulness. I have no idea how “Baxter” translates cross culturally, but I suspect that once on the job Baxter will be referred to by any number of personal names including the folksy “Bax”. It is clear that Rethink was purposeful in their selection of the name Baxter. They were making a point and it is well taken.

Conveying Intent

Rethink rightly understands that as an interface, it is difficult to improve upon the human face. After all, human faces as a means of communication have been proven in the field for over 200,000 years, millions of years if prehumans are considered. Developing a set of control icons that has the same universal, cross cultural appeal and understandability of a human face is a difficult, if not impossible, task.

Like a human, Baxter has many ways of conveying intent. Cues provided by the robot are relatively natural and no additional knowledge is required to use the robot as tasks and interactions are altered. For example, Baxter informs co-workers what it is about to do by looking in the direction where it will move its arm next. If a worker comes close to Baxter, the robot’s face exhibits a concerned expression. Using facial expressions Baxter can also inform you if it is confused or not working. Baxter’s interface is as simple, and as profound, as that.

Human Class, Human Scale, Human Pace

Baxter is humanoid in more than just its physical appearance. The system was designed to work at a pace and perform tasks in approximately the same way a human would. Processes requiring extremely rapid movement or exceedingly precise manipulation and placement are not Baxter’s forte. The same holds for hazardous or destructive tasks. These same undertakings cannot be accomplished by a human, nor can they be take place in close proximity with humans (and never without safety gear). Traditional industrial robots can perform these tasks, and it is for this reason that Rethink describes Baxter as a complement, and not a competitor, to traditional industrial robots.

Most industrial robotics systems automate the three D tasks–Dirty, Dull and Dangerous–with an emphasis on dirty and dangerous. By comparison, Rethink has optimized Baxter for automating the dull, those non-value add, repetitive tasks performed by humans in SMB manufacturing environments across the world. With Baxter, Rethink is conceding a good portion of the industrial manufacturing market to traditional industrial automation systems. It makes sense for Rethink to avoid direct competition with sizable, entrenched robotics companies. It could also be the simple matter of the company specifically targeting a much larger, underserved market. The reality, of course, is that both statements are true.

As a new market entrant, Rethink’s positioning is not unique in the least. What is unique is their contention that Baxter best serves SMB manufacturers when performing human class, human scale tasks in a human friendly environment. Baxter’s value proposition is that the robot can perform the tasks, allowing humans to focus on tasks that add greater value to the overall manufacturing process. Moreover, and here’s the kicker, if the robot goes offline for any reason, it can be replaced by a human, with no loss of production run downtime.

Political and Social Drivers

Virtually all of the major robotics producers are based outside of the US. Rethink notes that Baxter is designed in Boston and that approximately 70% of the value added sourcing occurs in the US. Final assembly takes place in New Hampshire. Rethink believes that this will resonate with manufactures in North America. While the company still has to makes it business case, they believe that Made in the USA can be a tie breaker in close competitive sales situations. Given today’s political and economic climate, they are most certainly correct.

More Than a Product Release

In an earlier article I speculated that Rethink’s product, when released, would be viewed as revolutionary and that the company would secure first-to-market benefits. With the Baxter announcement, it appears that these expectations are now confirmed.

Rethink Robotics’ much ballyhooed Baxter announcement is no point release or run-of-the-mill product launch. With the release of Baxter, the company is not offering a better industrial robot. Instead, the company has engineered a new category of manufacturing technology, an industrial partner robot, which is optimized for human scale, human paced manufacturing at small-to-medium companies. In doing so, Rethink Robotics will greatly expand the number and types of manufacturing tasks that can be automated. Indeed, the system will create wholly new classes of processes suitable for automation. Now that’s revolutionary.