February 10, 2016      

Joanne Pransky, associate editor of Industrial Robot, recently talked with Steve Cousins, the founder of Savioke. Cousins has a Ph.D. in computer science from Stanford University and led research at the Xerox Palo Alto Research Center and the IBM Almaden Research Center.

Cousins also oversaw the creation of the Robot Operating System, the PR2 robot, and the open-source TurtleBot at Willow Garage. In addition to spinning off several robotics companies from Willow Garage, he is a pioneer in the robots as a service (RaaS) business model with Savioke‘s Relay robot in hotels. Cousins shares some of experiences as a business and technology leader.

This interview is available free to Robotics Business Review readers until March 15, 2016.

Pransky: Could you please explain the technical evolution from how you got from the mobile manipulators of PR2 to the current design of Relay? And do you plan to add manipulators at some point in the future?

Cousins: We found a strategy of “minimum viable product.” It’s a very popular strategy among start-ups in which you build the minimum product that’s viable and don’t do anything more than you have to in order to get something out the door.

A PR2 robot at the University of California, Berkeley

A PR2 robot at UC Berkeley masters motor tasks via a deep-learning algorithm.

Even at Willow Garage, we went from a PR2 with two seven-degree-of-freedom arms, designed for a much broader set of tasks to basically move around human environments to perform human tasks, down to a one-armed robot called Platform Bot that spun off as Unbounded Robotics.

We knew when we started Savioke that it was going to be very expensive to build something like the PR2. We actually continued on the simplifying path and said, “Do we even need one humanoid-like arm? Or is that overkill?”

Just because we humans have arms doesn’t mean that that’s the only way objects can be manipulated.

What we found at Savioke were enough use cases where somebody’s around at the robot loading and unloading time and therefore an item doesn’t need to be picked up automatically. That means that you don’t have to even have that manipulation.

It was really a needs analysis driving the function as opposed to saying, “Why don’t we just add these arbitrary manipulators?” We could add anything, but we only put materials in that have a value.

Pransky: Do you see intellectual property (IP) as a good thing or a bad thing for start-up companies? In other words, is it essential for survival, or can it impede ideas and freethinking and restrict spin-offs?

Cousins: I would say that intellectual property is necessary for venture-backed start-up companies. It’s very difficult to raise money unless you can answer why it is that you have an advantage of some kind.

Of course, you don’t have to have IP. You could have particularly strong connections with the market that give you a key advantage. But you have to show some kind of advantage position, such as, why is this team going to succeed here? “Why doesn’t someone else come along and eat your lunch?” is a question that every investor that you talk to asks. Thus, you have to do something.

But at the same time, intellectual property is friction. You can spend a lot trying to negotiate intellectual property deals. You can give it away. You can spend a long time litigating. You have to be careful about it. Intellectual property is not just patents. There’s a wide range of intellectual property that includes the artifacts that you build, the source code, the know-how, the business plans, etc.

At Willow Garage, we had the nice position that we could afford to be very open and make a lot of open-source contributions. In fact, that was kind of the philosophy there which gave us huge advantages in terms of recruiting and in terms of disseminating [the Robot Operating System].

ROS took off because it was open-source. If your goal is to make a standard, open source is a great way to go. But if your goal is to raise venture capital, you have to figure out, what is my advantage with that?

Pransky: In 10 years’ time, do you expect crowdfunding to be a major player or a failed experiment?

Cousins: I think crowdfunding has a very big place. I don’t know whether it’s going to be either a failed experiment or a major player. People use crowdfunding to test interest in a product, especially a new interest in a product in a way that’s hard to do otherwise.

Jibo, as an example, was very successful because they got a lot of crowdfunding and a lot of interest upfront early. There are a number of examples similar to that. If you’re going directly to consumers, crowdfunding is interesting.

We’ve thought about crowdfunding for Savioke, but as a business, we’re not likely to get hotels to contribute and to play the crowdfunding game. It’s something that’s really designed to the individual consumer. I don’t think it’s the way that you’re going to fund a serious development effort in most cases.

Pransky: What do you think Ph.D. and Master of Engineering students should be doing while in school to best prepare them for the commercial side of robotics?

Cousins: I think it’s helpful to take the business classes when they’re offered. I took a class called “Business Management for Computer Scientists” at Stanford. We went through Harvard Business School studies and strategies. That was very helpful to me down the road.

At the time, I didn’t do a lot with it, but I think for anyone who’s entrepreneurial, it’s worth taking some business classes and learning the rudiments to understand and to see if you like the business aspects.

Click here to read the entire interview.

More Essential Interviews: