Rethink Robotics’ cobots were iconic, but what robot startup lessons can we take from its failure? We spoke with several industry experts to find out.
Last week, Rethink Robotics Inc.’s announcement that it had shut down came as a surprise to many people. However, others in the automation industry recognized the company’s struggles as an opportunity to review robot startup lessons.
Boston-based Rethink Robotics sold the Baxter and Sawyer collaborative robot arms. Its cobots were well-known, particularly among researchers, but they apparently failed to find enough customers in manufacturing.
Sad but not surprised
Rethink‘s demise was felt hard in the close-knit community around Boston, which is home to more than 150 robotics companies.
“Rethink has been a great partner to MassRobotics, and we were sad to hear the news,” said Tom Ryden, executive director of MassRobotics, an organization devoted to nurturing the Massachusetts robotics ecosystem. “Rethink has pioneered technology innovations for workplace robots that will forever change the way people deploy robots in manufacturing and logistics environments.”
“It’s always a surprise to hear news like this, particularly when it’s about an industry pioneer like Rethink, a company that essentially invented the collaborative robot category,” said Rick Faulk, CEO of mobile robot maker Locus Robotics.
“I was saddened to hear that Rethink Robotics had closed its doors, but I wasn’t surprised,” said Prasad Akella, founder and CEO of Drishti, a startup working on applying artificial intelligence and “action recognition to digitize human activities” for greater efficiency.
“Building a successful startup isn’t easy — Drishti is my second company, and it’s a tough road,” he stated. “Compounding the typical startup challenges, Rethink Robotics was also a pioneer proving a new technology and creating a category.”
“I did something similar at my last company, Spoke Software, and now again at Drishti, so I can attest to how hard it is to get buy-in from an audience that has never considered your solution before because it didn’t exist until you showed up,” Akella said.
More robot startup lessons
Other executives agreed that the growth of robotics shouldn’t be taken for granted and that one of the most important robot startup lessons is to be unafraid of failure.
In addition to Ryden, Faulk, and Akella, Robotics Business Review spoke with several industry leaders, including the following:
- Jeff Burnstein, president of the Association for Advancing Automation (A3), the parent organization of the Robotic Industries Association, among others
- Peter Harris, CEO of HighRes Biosolutions, a provider of laboratory automation for drug discovery and scientific research
- Vince Martinelli, head of product and marketing at RightHand Robotics Inc., which makes picking hardware and software for the supply chain industry
- John Santagate, research director of commercial service robotics at analyst firm IDC
Here are their responses:
Q: Why did Rethink Robotics fail?
Harris: Rodney Brooks and Jim Lawton did a really important service with starting the conversation around cobots and robot design. They pushed the idea of humans and robots working together in shared spaces, but they didn’t match it with competitive technology.
We knew they were exploring strategic alternatives, but it’s always abrupt when a company closes its doors.
There are more than two tiers of collaborative robots. Baxter and Sawyer were iconic, but it always made more sense to go with a UR5. Universal Robots’ robots are significantly better than Rethink’s cobots.
The core performance parameters are reasonably straightforward: speed, precision, repeatability, accuracy, and overall smoothness of motion. When you evaluate a Baxter and an iiwa [from KUKA AG] — how it moves through its motion path — it’s not even a close comparison.
Beyond a robust design and construction for high-performance applications, you need to evaluate the control system and how easy parts are to repair. In a UR arm, you can pull out a piece and replace it. That’s a better piece of industrial engineering.
Baxter was this eyeball-faced robot that drove the conversation, but it was not practically interesting to people who needed to perform stuff. We’d never put something like that on a laboratory floor.
Q: Is this a sign of trouble for automation and cobots specifically, or does it reflect the challenges facing new companies?
Burnstein: Rethink’s situation is not a sign of trouble for automation or collaborative robots. In fact, the entire robotics and automation market is booming globally, fueled by interest in cobots. Universal Robots recently announced the sale of its 25,000th robot.
Many new companies are entering this space, and we’re expecting a large turnout at our Collaborative Robots, Advanced Vision, and AI Conference in Santa Clara, Calif., on Oct. 24 and 25.
Faulk: While there’s been an explosive growth in the market, in order to be successful, collaborative robots need to be able to deliver hardened economic value propositions. And, as robotic technology advances, these technologies must solve real problems and address specific operational needs. A company can’t simply rely on the advancing technology to be successful.
Martinelli: I think every company’s journey is unique, but in this case, it may be related more to the challenges that new companies must face. For example, there were several thousand companies making cars at the dawn of the automobile age in the late 19th century; by the 1920s, there were only three in the U.S.
Santagate: I think not. I think it is more of a function of the risk of being a pioneer in a new space. Rethink helped to create the market for collaborative robotics that are safe, easy to use, and extremely flexible.
However, when they entered the market, they made a bit of a splash, and other more established robotic vendors in the industrial robotics space took notice and came out with their own slant on collaborative robotics.
Companies like ABB, Yaskawa, KUKA, and FANUC have extended their portfolios to include user-friendly, highly capable, and very flexible collaborative robotics.
We also have to consider Universal Robots, the leader in terms of collaborative robotic arms. They have been around since 2005 and have consistently improved their products in terms of capability and usability. UR has done so rather cost-effectively.
I think the market for cobot arms will continue to increase, but the risk is there for emerging entrants. The bigger players have the existing customer, distribution, and partner networks, as well as the financial resources to win in the space. For new entrants, it will be quite difficult, as these major players are constantly innovating in this space now.
Harris: I don’t think this is a sign of trouble at the aggregate level. Lots of people want to grab a piece of the pie, but not everyone will make it. Success is unevenly distributed across players; it’s the natural way all industries evolve over time.
Akella: Automation is here to stay. It’s going to modify every industry and an enormous number of tasks over the next decade.
Rethink’s departure verifies a couple of truths that we already suspected: First, robot startups need to be hyper-focused on solving true manufacturing problems and demonstrating ROI quickly. It’s tempting to get caught up in the newness and promise of robotics, but you have to ensure the developments you’re making with the cobot lead to measurable gains for the plant manager.
Second, humans are still very much a part of the manufacturing environment, and any steps toward automation and digitization need to include tasks done by humans, or they’re incomplete.
Q: What robot startup lessons can you share regarding matching technologies with markets?
Burnstein: My guess is that every robot startup faces a minefield of issues relating to product, market timing, managing investor expectations, etc. Rethink played an important role in establishing the viability of this new type of robot that could work side by side with people.
Faulk: The industry needs to continually be looking for the next innovation to advance collaborative technology, but the innovation itself needs to solve real-world needs in a cost-effective manner.
Martinelli: It’s always critical for startups to solve real business problems for customers to create market pull for the solution they provide. The sooner a B2B technology startup can get their solution in the hands of paying customers, the sooner they can learn what will succeed and what won’t. Although filling a need is your friend, time is the enemy.
Santagate: First off, just because a company can deploy automation doesn’t mean that it should. It is not enough to create a new robotic product unless it solves for some business process at scale. Companies looking to enter the market should take into account several things:
- Is the product being designed to solve a problem, or is it a solution that will be looking for a problem? The latter is rarely successful, while the former delivers success.
- Is there a current competitive environment that can adapt and deliver a similar solution, in short order, and have an existing network by which to exploit to drive market share? If there is, is there IP [intellectual property] that can help protect the innovation and the market opportunity? If there is an existing market, what is the exit strategy? Define this early on and configure the business to support this exit strategy.
- Look for scale. Find a market opportunity that has sufficient scale to support growth, and ensure that product development and organizational structure is sufficient to meet the demands of that scale.
Akella: Startups in any industry need to fully understand their market, the challenges their buyers face, and how to solve real problems with their technology — the “product-market fit,” as VCs like to point out.
In the case of Rethink, it seems that there were some clear tradeoffs in terms of functionality early on that compromised the usefulness of the robots down the road. For example, the robots were safe to operate around humans, but the payloads were pretty small, and the positioning precision was compromised.
This may have made it difficult for manufacturers to see real value in their cobot purchase, which limited the market for Rethink. Manufacturers need to see genuine improvements in terms of productivity and quality before they’ll purchase new technology at scale.
Q: How are cobots evolving?
Harris: Rethink’s demise is a good reminder that it’s really quite likely that the destination for collaborative robots will remerge with industrial robotics.
As the technologies for sensing and environmental management improve, this notion that they’re two distinct universes will dissolve. With companies like KUKA and Denso, few can afford to be a niche cobot maker. The more successful robotics companies over time will have to have a blended portfolio.
We’re still a ways away from an industrial robot operating in a true collaborative mode. This isn’t lost on the big robot makers. They’re going to spend time in the brackish water between industrial and collaborative robots.
Cobot makers can’t overspecialize, and Universal Robots’ support ecosystem is a natural evolution. It’s impressive that they’ve expanded with as narrow a focus as they’ve had, and parent company Teradyne, which also bought Mobile Industrial Robots, helps with that.
Q: Should robot startups expect cutthroat competition?
Harris: The robotics business is a tough business. People have this glorified sense that it’s just this giant rising tide, lifting everyone who has “robot” in their business description.
The cobot arm business is incredibly competitive. The technology differences are not very large, and the larger entrenched arm players are pushing their differentiators so the incremental cost to catch up is incredibly high.
Q: Do you see a similar shakeout coming as multiple companies try to get into the mobile robotics space?
Burnstein: The mobile robot space is certainly interesting, with many companies pursuing the large opportunities that exist in part due to the explosion of e-commerce.
Will some of the entrants into this space fail? Will there be acquisitions within the space? I think it’s too soon to draw conclusions, and certainly not based on Rethink’s situation.
Faulk: The industry is growing quickly, with the introduction of many new technologies offering multiple ways to “solve” a similar problem. This growth is a reflection of the diverse challenges that warehouse and manufacturing companies are addressing for their products and customers.
At Locus, we see significant opportunity for several of these technologies to be used to solve specific needs versus a one-size-fits-all approach.
Martinelli: It’s too early to say. There seem to be many opportunities for a range of solutions, depending on the end-user business model, geography, etc. Each company in that space needs to stay focused on what they can do for their customers.
Santagate: I have been saying this exact thing for nearly a year now. I do believe that there will be a bit of a shakeout coming in the autonomous mobile robot [AMR] space. There are over 20 vendors with a product built for AMR operations in the warehouse and fulfillment center.
There is a cataclysmic shakeout coming for mobile robots.
—Peter Harris, CEO, HighRes Biosolutions
Now, while this market is sufficient to support an overabundance of warehouse management software [WMS] applications, it is not enough to support a hardware-driven business.
Software companies can successfully operate with a micro-vertical approach. Companies building complex hardware-centric technology combined with software have a different cost structure that makes it much harder to thrive with limited market share.
I believe we will see several more acquisitions in the space, and we’ll also see a bit of attrition. There are several vendors that are making big noise in the space and are capturing or are planning to capture significant market share.
Others are very early on [in development and commercialization and] looking to break in. I think those with a purpose-built approach will likely thrive.
Harris: There is a cataclysmic shakeout coming for mobile robots.
If evolution was causing crowding in the arm space, imagine the hardware commoditization in all these carts — 50 bazillion people are making small indoor vehicles.
This lesson really turns on which companies can be not just a small mobile robot maker. They have to be integrated solution providers, where the software suite is about way more than the robot. Any kid can build a mobile robot; they’ll all come from China soon.
An enterprise solution that solves a problem can’t be duplicated very well. The world is focused on a robot startup’s cute bit of hardware, but for large warehouses and logistics operations, it’s all about how to take series of nodes and solve the traveling salesman problem.
They can’t all get bought like Amazon did with Kiva, not at the valuations they’re hoping for. It’s an interesting question: How much of this becomes captive like Amazon Robotics, or a consolidated third-party provider? DHL could buy one of these robot startups.
It’s whether the piece parts just become components that significant end users integrate themselves or leave to someone like Locus to manage it for them.
In warehouse and logistics, the irreplaceable thing that every end user lives and dies by is the ERP system. Companies have to ask, “How much differentiation do I need with systems sending and receiving to robots and the task or schedule manager?”
Akella: We’ve all seen the mobile robotics space explode in the last few years. Mobile robots are simpler than cobots — they don’t require as much precision or adaptability — and therefore, they’ll likely become commoditized much quicker than cobots.
But the bigger problem that I keep hearing about is the fact that these are still fundamentally fancy tugs. It’s no surprise, therefore, that Amazon is now the largest employer in the U.S. All those warehouses that they are building need large numbers of humans to do the picking and sorting and the kitting that robots still can’t do!
So, my sense is that the space has been overblown in terms of its capabilities, and reality will soon set in. We all agree that what mobile robots do is limited, but useful. We need more innovation to really solve the warehousing problems of Walmart and 7-Eleven, as examples.