Robotic Dreams, Robotic Realities: Why Is It So Hard to Build Profitable Robot Companies?

Roboticists need to discuss openly and honestly not only our successes but also our failures

3 min read
Sad Kuri robot
Mayfield Robotics, the company behind Kuri, announced last year that the social home robot had been canceled.
Image: IEEE Spectrum

A version of this article appears in the IEEE Robotics & Automation Magazine (Volume 26, Issue 1, March 2019).

In mid-November, we received the sad news that Alphabet is closing SCHAFT, a spinoff of the University of Tokyo robotics lab. The decision comes one year after Boston Dynamics was sold to SoftBank, the company that also acquired Aldebaran Robotics (known for the Pepper and Nao robots). During the 2018 IEEE/Robotics Society of Japan International Conference on Intelligent Robots and Systems, we heard that Rethink Robotics, which created the collaborative robot industry and had a large impact on our view of robots in industrial applications, had closed its doors.

Some months before, Jibo and Mayfield Robotics, makers of Kuri, were forced to shut down sales and operations. Jibo was once heralded as “the first social robot for the home” and was named one of Time’s “Best Inventions of 2017.” Other than a few robot vacuum companies (mainly iRobot), no company has developed a successful home robot.

Sad Jibo Late last year, Jibo shut down its Boston office and completed the sale of its assets and intellectual property to a New York–based investment management firm. Image: IEEE Spectrum

The news initiated a discussion on Facebook among robotics leaders, such as Chris Atkeson from Carnegie Mellon University’s Robotics Institute; James Kuffner, chief executive officer of the Toyota Research Institute - Advanced Development (TRI-AD); and Giulio Sandini from the Italian Institute of Technology. All agree the robotics industry is still on the rise; it is just extremely hard to make a profitable robotics company. But, unless big bets are made, new research and technology will never mature into products that are practical and useful for the world. Moreover, success in this area demands more than good technology. As James Kuffner stated, “It requires significant funding, committed leadership, highly skilled staff, resources, and infrastructure, and an excellent product and market strategy. Not to mention flawless execution. It is unrealistic to expect every effort to succeed.”

Overselling is a dangerous strategy that can be counterproductive. Both companies and researchers publish videos of robots doing tasks, but sometimes they fail to point out the limitations of the technology or that those results were achieved in lab conditions.

Chris Atkeson raised the big questions: What have we learned from the failures? How can we further build on the work? What lessons can be taken? How will the intellectual property be transferred? The future will tell whether the know-how will be reincarnated. Often, the work is secret, especially when sponsored by the military, and only amazing YouTube videos are released. However, some companies choose to contribute to the open source movement, with the Robotic Operating System (ROS) as probably the most well-known example.

Moreover, the investments of tech giants in robotics and artificial intelligence energized and catalyzed the industry, resulting in billions of dollars of additional investment in research and development around the world. Hopefully, companies will also publish—e.g., in IEEE journals and magazines—more scientific insights on their products.

The problem, as Giulio Sandini put it, occurs when one sells (or buys) intentions as results. Overselling is a dangerous strategy that can be counterproductive, even for the whole robotics community. Both companies and researchers publish videos of robots doing tasks, but sometimes they fail to point out the limitations of the technology or that those results were achieved in lab conditions. This makes it much more difficult to explain to nonroboticist industry executives the difference between creating a one-off demo and creating a real product that works reliably.

Deep learning, for example, is at the forefront of the AI revolution, but it is too often viewed as the magic train carrying us into the world of technological wonders. AI researchers are warning about overexcitement and that the next AI winter is coming.

The first cracks are already visible, as is the case of the promises claimed for self-driving cars. Rodney Brooks, founder of Rethink Robotics, regularly writes relevant essays on this topic on his blog. Robot ethics professor Noel Sharkey wrote an article in Forbes titled “Mama Mia It’s Sophia: A Show Robot or Dangerous Platform to Mislead?Tony Belpaeme, a social robot researcher from the University of Ghent, replied with a tweet, “I had [a European Union] project reviewer express disappointment in our slow research progress, as the Sophia bot clearly showed that the technical challenges we were still struggling with were solved.”

It is our common responsibility and interest to disseminate openly and honestly not only our success but also our failures. Together, we can realize our dreams for numerous robotic applications and devise a realistic plan to develop them.

Bram Vanderborght is a professor at the Brussels Human Robotics Research Center, part of Vrije Universiteit Brussel, in Belgium, and the editor in chief of IEEE Robotics & Automation Magazine. His research interests include cognitive and physical human-robot interaction, robot assisted therapy, humanoids, and rehabilitation robotics.

The Conversation (0)

How the U.S. Army Is Turning Robots Into Team Players

Engineers battle the limits of deep learning for battlefield bots

11 min read
Robot with threads near a fallen branch

RoMan, the Army Research Laboratory's robotic manipulator, considers the best way to grasp and move a tree branch at the Adelphi Laboratory Center, in Maryland.

Evan Ackerman

“I should probably not be standing this close," I think to myself, as the robot slowly approaches a large tree branch on the floor in front of me. It's not the size of the branch that makes me nervous—it's that the robot is operating autonomously, and that while I know what it's supposed to do, I'm not entirely sure what it will do. If everything works the way the roboticists at the U.S. Army Research Laboratory (ARL) in Adelphi, Md., expect, the robot will identify the branch, grasp it, and drag it out of the way. These folks know what they're doing, but I've spent enough time around robots that I take a small step backwards anyway.

This article is part of our special report on AI, “The Great AI Reckoning.”

The robot, named RoMan, for Robotic Manipulator, is about the size of a large lawn mower, with a tracked base that helps it handle most kinds of terrain. At the front, it has a squat torso equipped with cameras and depth sensors, as well as a pair of arms that were harvested from a prototype disaster-response robot originally developed at NASA's Jet Propulsion Laboratory for a DARPA robotics competition. RoMan's job today is roadway clearing, a multistep task that ARL wants the robot to complete as autonomously as possible. Instead of instructing the robot to grasp specific objects in specific ways and move them to specific places, the operators tell RoMan to "go clear a path." It's then up to the robot to make all the decisions necessary to achieve that objective.

Keep Reading ↓ Show less